Showing 1–11 of 11 results for author: Whitmer, J K

Considerations in the use of ML interaction potentials for free energy calculations

Authors: Orlando A. Mendible, Jonathan K. Whitmer, Yamil J. Colón

Abstract: Machine learning force fields (MLFFs) promise to accurately describe the potential energy surface of molecules at the ab initio level of theory with improved computational efficiency. Within MLFFs, equivariant graph neural networks (EQNNs) have shown great promise in accuracy and performance and are the focus of this work. The capability of EQNNs to recover free energy surfaces (FES) remains to be… ▽ More Machine learning force fields (MLFFs) promise to accurately describe the potential energy surface of molecules at the ab initio level of theory with improved computational efficiency. Within MLFFs, equivariant graph neural networks (EQNNs) have shown great promise in accuracy and performance and are the focus of this work. The capability of EQNNs to recover free energy surfaces (FES) remains to be thoroughly investigated. In this work, we investigate the impact of collective variables (CVs) distribution within the training data on the accuracy of EQNNs predicting the FES of butane and alanine dipeptide (ADP). A generalizable workflow is presented in which training configurations are generated with classical molecular dynamics simulations, and energies and forces are obtained with ab initio calculations. We evaluate how bond and angle constraints in the training data influence the accuracy of EQNN force fields in reproducing the FES of the molecules at both classical and ab initio levels of theory. Results indicate that the model's accuracy is unaffected by the distribution of sampled CVs during training, given that the training data includes configurations from characteristic regions of the system's FES. However, when the training data is obtained from classical simulations, the EQNN struggles to extrapolate the free energy for configurations with high free energy. In contrast, models trained with the same configurations on ab initio data show improved extrapolation accuracy. The findings underscore the difficulties in creating a comprehensive training dataset for EQNNs to predict FESs and highlight the importance of prior knowledge of the system's FES. △ Less

Submitted 14 May, 2025; v1 submitted 20 March, 2024; originally announced March 2024.

PySAGES: flexible, advanced sampling methods accelerated with GPUs

Authors: Pablo F. Zubieta Rico, Ludwig Schneider, Gustavo R. Pérez-Lemus, Riccardo Alessandri, Siva Dasetty, Cintia A. Menéndez, Yiheng Wu, Yezhi Jin, Yinan Xu, Trung D. Nguyen, John A. Parker, Andrew L. Ferguson, Jonathan K. Whitmer, Juan J. de Pablo

Abstract: Molecular simulations are an important tool for research in physics, chemistry, and biology. The capabilities of simulations can be greatly expanded by providing access to advanced sampling methods and techniques that permit calculation of the relevant underlying free energy landscapes. In this sense, software that can be seamlessly adapted to a broad range of complex systems is essential. Buildin… ▽ More Molecular simulations are an important tool for research in physics, chemistry, and biology. The capabilities of simulations can be greatly expanded by providing access to advanced sampling methods and techniques that permit calculation of the relevant underlying free energy landscapes. In this sense, software that can be seamlessly adapted to a broad range of complex systems is essential. Building on past efforts to provide open-source community supported software for advanced sampling, we introduce PySAGES, a Python implementation of the Software Suite for Advanced General Ensemble Simulations (SSAGES) that provides full GPU support for massively parallel applications of enhanced sampling methods such as adaptive biasing forces, harmonic bias, or forward flux sampling in the context of molecular dynamics simulations. By providing an intuitive interface that facilitates the management of a system's configuration, the inclusion of new collective variables, and the implementation of sophisticated free energy-based sampling methods, the PySAGES library serves as a general platform for the development and implementation of emerging simulation techniques. The capabilities, core features, and computational performance of this new tool are demonstrated with clear and concise examples pertaining to different classes of molecular systems. We anticipate that PySAGES will provide the scientific community with a robust and easily accessible platform to accelerate simulations, improve sampling, and enable facile estimation of free energies for a wide range of materials and processes. △ Less

Submitted 4 April, 2023; v1 submitted 12 January, 2023; originally announced January 2023.

Transfer Learning Facilitates the Prediction of Polymer-Surface Adhesion Strength

Authors: Jiale Shi, Fahed Albreiki, Yamil J. Colón, Samanvaya Srivastava, Jonathan K. Whitmer

Abstract: Machine learning (ML) accelerates the exploration of material properties and their links to the structure of the underlying molecules. In previous work [J. Shi, M. J. Quevillon, P. H. A. Valença, and J. K. Whitmer, \textit{ACS Appl. Mater. Interfaces.}, 2022, 14, 32, 37161--37169], ML models were applied to predict the adhesive free energy of polymer--surface interactions with high accuracy from t… ▽ More Machine learning (ML) accelerates the exploration of material properties and their links to the structure of the underlying molecules. In previous work [J. Shi, M. J. Quevillon, P. H. A. Valença, and J. K. Whitmer, \textit{ACS Appl. Mater. Interfaces.}, 2022, 14, 32, 37161--37169], ML models were applied to predict the adhesive free energy of polymer--surface interactions with high accuracy from the knowledge of the sequence data, demonstrating successes in inverse-design of polymer sequence for known surface compositions. While the method was shown to be successful in designing polymers for a known surface, extensive datasets were needed for each specific surface in order to train the surrogate models. Ideally, one should be able to infer information about similar surfaces without having to regenerate a full complement of adhesion data for each new case. In the current work, we demonstrate a transfer learning (TL) technique using a deep neural network to improve the accuracy of ML models trained on small datasets by pre-training on a larger database from a related system and fine-tuning the weights of all layers with a small amount of additional data. The shared knowledge from the pre-trained model facilitates the prediction accuracy significantly on small datasets. We also explore the limits of database size on accuracy and the optimal tuning of network architecture and parameters for our learning tasks. While applied to a relatively simple coarse-grained (CG) polymer model, the general lessons of this study apply to detailed modeling studies and the broader problems of inverse materials design. △ Less

Submitted 5 January, 2023; originally announced January 2023.

Free Energy Landscape and Isomerization Rates of Au$_4$ Clusters at Finite Temperature

Authors: Jiale Shi, Shanghui Huang, François Gygi, Jonathan K. Whitmer

Abstract: In metallic nanoparticles, the cluster geometric structures control the particle's electronic band structure, polarizability, and catalytic properties. Analyzing the structural properties is a complex problem; the structure of an assembled cluster changes from moment to moment due to thermal fluctuations. Conventional structural analyses based on spectroscopy or diffraction cannot determine the in… ▽ More In metallic nanoparticles, the cluster geometric structures control the particle's electronic band structure, polarizability, and catalytic properties. Analyzing the structural properties is a complex problem; the structure of an assembled cluster changes from moment to moment due to thermal fluctuations. Conventional structural analyses based on spectroscopy or diffraction cannot determine the instantaneous structure exactly and can merely provide an averaged structure. Molecular simulations offer an opportunity to examine the assembly and evolution of metallic clusters, as the preferred assemblies and conformations can easily be visualized and explored. Here, we utilize the adaptive biasing force algorithm applied to first principles molecular dynamics to demonstrate exploration of a relatively simple system which permits comprehensive study of the small metal cluster $\ce{Au4}$ in both neutral and charged configurations. Our simulation work offers a quantitative understanding of these clusters' dynamic structure, which is significant for single-site catalytic reactions on metal clusters and provides a starting point for a detailed quantitative understanding of more complex pure metal and alloy clusters' dynamic properties. △ Less

Submitted 11 October, 2021; originally announced October 2021.

Comments: This work was supported by MICCoM, the Midwest Center for Computational Materials (www.miccom-center.org) s part of the Computational Materials Sciences Program funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division

Predicting Adhesive Free Energies of Polymer--Surface Interactions with Machine Learning

Authors: Jiale Shi, Michael J. Quevillon, Pedro H. Amorim Valença, Jonathan K. Whitmer

Abstract: Polymer-surface interactions are crucial to many biological processes and industrial applications. Here we propose a machine-learning method to connect a model polymer's sequence with its adhesion to decorated surfaces. We simulate the adhesive free energies of $20,000$ unique coarse-grained 1D sequential polymers interacting with functionalized surfaces and build support vector regression (SVR) m… ▽ More Polymer-surface interactions are crucial to many biological processes and industrial applications. Here we propose a machine-learning method to connect a model polymer's sequence with its adhesion to decorated surfaces. We simulate the adhesive free energies of $20,000$ unique coarse-grained 1D sequential polymers interacting with functionalized surfaces and build support vector regression (SVR) models that demonstrate inexpensive and reliable prediction of the adhesive free energy as a function of the sequence. Our work highlights the promising integration of coarse-grained simulation with data-driven machine learning methods for the design of new functional polymers and represents an important step toward linking polymer compositions with polymer-surface interactions. △ Less

Submitted 6 October, 2021; originally announced October 2021.

Exploring the Potential of Parallel Biasing in Flat Histogram Methods

Authors: Shanghui Huang, Michael J. Quevillon, Ernesto C. Cortés-Morales, Jonathan K. Whitmer

Abstract: Metadynamics, a member of the `flat histogram' class of advanced sampling algorithms, has been widely used in molecular simulations to drive the exploration of states separated by high free energy barriers and promote comprehensive sampling of free energy landscapes defined on collective variables (CVs) which characterize the state of the system. Typically, the methods encounter severe limitations… ▽ More Metadynamics, a member of the `flat histogram' class of advanced sampling algorithms, has been widely used in molecular simulations to drive the exploration of states separated by high free energy barriers and promote comprehensive sampling of free energy landscapes defined on collective variables (CVs) which characterize the state of the system. Typically, the methods encounter severe limitations when exploring large numbers of CVs. A recently proposed variant, parallel bias metadynamics (PBMetaD), promises to aid in exploring free energy landscapes along with multiple important collective variables by exchanging the $n$-dimensional free energy landscape required by standard methods for $n$ one-dimensional marginal free energy landscapes. In this study, we systematically examine how parallel biasing affects the convergence of free energy landscapes along with each variable relative to standard methods and the effectiveness of the parallel biasing strategy for addressing common bottlenecks in the use of advanced sampling to calculate free energies. △ Less

Submitted 10 September, 2021; originally announced September 2021.

Comments: SH and JKW acknowledge support for this project from the United States National Science Foundation (Award No. DMR-1751988). MJQ and ECM were supported by the MICCoM Center at Argonne National Laboratory under a Computational Materials Science center grant from the Department of Energy, Basic Energy Sciences Division

Linking dynamics and structure in highly asymmetric ionic liquids

Authors: Mariana E. Farías-Anguiano, Ernesto C. Cortés-Morales, Jonathan K. Whitmer, Pedro E. Ramírez-González

Abstract: We explore an idealized theoretical model for the transport of ions within highly asymmetric ionic liquid mixtures. A primitive model (PM)-inspired system serves as a representative for asymmetric ionic materials (such as liquid crystalline salts) which quench to form disordered, partially-arrested phases. Self-Consistent Generalized Langevin Equation (SCGLE) Theory is applied to understand the co… ▽ More We explore an idealized theoretical model for the transport of ions within highly asymmetric ionic liquid mixtures. A primitive model (PM)-inspired system serves as a representative for asymmetric ionic materials (such as liquid crystalline salts) which quench to form disordered, partially-arrested phases. Self-Consistent Generalized Langevin Equation (SCGLE) Theory is applied to understand the connection between the size ratio of charge-matched salts and their average mobility. Within this model, we identify novel glassy states where one of the two charged species (either the macro-cation or the micro-anion) are arrested, while the other retains mobility. We discuss how this result is useful in the development of novel single-ion conducting phases in ionic liquid based materials. △ Less

Submitted 27 August, 2021; originally announced August 2021.

Comments: 11 pages, 5 figures

Automatic Determination of $n$-Cyanobiphenyl Elastic Constants from Molecular Simulation

Authors: Hythem Sidky, Jonathan K. Whitmer

Abstract: New applications of liquid crystalline materials have increased the need for precise engineering of elastic properties. Recently, Sidky et al. presented methods by which the elastic coefficients of molecular models with atomistic detail can be accurately calculated, demonstrating the result for the ubiquitous mesogen 5CB. In this work, these techniques are applied to the homologous series of nCB m… ▽ More New applications of liquid crystalline materials have increased the need for precise engineering of elastic properties. Recently, Sidky et al. presented methods by which the elastic coefficients of molecular models with atomistic detail can be accurately calculated, demonstrating the result for the ubiquitous mesogen 5CB. In this work, these techniques are applied to the homologous series of nCB materials, focusing on the standard bend, twist, and splay deformations, using an entirely automated process. Our results show strong agreement with published experimental measurements for the nCBs and present a path forward to computational molecular engineering of liquid crystal elasticity for novel molecules and mixtures. △ Less

Submitted 28 February, 2019; originally announced February 2019.

Comments: 6 pages, 4 figures. All scripts and information necessary to run the examples contained in this article are posted at https://github.com/hsidky/atomistic_elastics

Learning Free Energy Landscapes Using Artificial Neural Networks

Authors: Hythem Sidky, Jonathan K. Whitmer

Abstract: Existing adaptive bias techniques, which seek to estimate free energies and physical properties from molecular simulations, are limited by their reliance on fixed kernels or basis sets which hinder their ability to efficiently conform to varied free energy landscapes. Further, user-specified parameters are in general non-intuitive, yet significantly affect the convergence rate and accuracy of the… ▽ More Existing adaptive bias techniques, which seek to estimate free energies and physical properties from molecular simulations, are limited by their reliance on fixed kernels or basis sets which hinder their ability to efficiently conform to varied free energy landscapes. Further, user-specified parameters are in general non-intuitive, yet significantly affect the convergence rate and accuracy of the free energy estimate. Here we propose a novel method wherein artificial neural networks (ANNs) are used to develop an adaptive biasing potential which learns free energy landscapes. We demonstrate that this method is capable of rapidly adapting to complex free energy landscapes and is not prone to boundary or oscillation problems. The method is made robust to hyperparameters and overfitting through Bayesian regularization which penalizes network weights and auto-regulates the number of effective parameters in the network. ANN sampling represents a promising innovative approach which can resolve complex free energy landscapes in less time than conventional approaches while requiring minimal user input. △ Less

Submitted 7 December, 2017; originally announced December 2017.

Comments: 10 pages, 7 figures

Orientationally Glassy Crystals of Janus Spheres

Authors: Shan Jiang, Jing Yan, Jonathan K. Whitmer, Stephen M. Anthony, Erik Luijten, Steve Granick

Abstract: Colloidal Janus spheres in water (one hemisphere attractive and the other repulsive) assemble into two-dimensional hexagonal crystals with orientational order controlled by anisotropic interactions. We exploit the decoupled translational and rotational order to quantify the orientational dynamics. Via imaging experiments and Monte Carlo simulations we demonstrate that the correlations in the orien… ▽ More Colloidal Janus spheres in water (one hemisphere attractive and the other repulsive) assemble into two-dimensional hexagonal crystals with orientational order controlled by anisotropic interactions. We exploit the decoupled translational and rotational order to quantify the orientational dynamics. Via imaging experiments and Monte Carlo simulations we demonstrate that the correlations in the orientation of individual Janus spheres exhibit glasslike dynamics that can be controlled via the ionic strength. Thus, these colloidal building blocks provide a particularly suitable model glass system for elucidating nontrivial dynamics arising from directional interactions, not captured by the consideration of just translational order. △ Less

Submitted 10 June, 2014; originally announced June 2014.

Journal ref: Phys. Rev. Lett. 112, 218301 (2014)

Coarse-grained Modeling of DNA Curvature

Authors: Gordon S. Freeman, Daniel M. Hinckley, Joshua P. Lequieu, Jonathan K. Whitmer, Juan J. de Pablo

Abstract: Modeling of DNA-protein interactions is a complex process involving many important time and length scales. This can be facilitated through the use of coarse-grained models which reduce the number of degrees of freedom and allow efficient exploration of binding configurations. It is known that the local structure of DNA can significantly affect its protein-binding properties (i.e. intrinsic curvatu… ▽ More Modeling of DNA-protein interactions is a complex process involving many important time and length scales. This can be facilitated through the use of coarse-grained models which reduce the number of degrees of freedom and allow efficient exploration of binding configurations. It is known that the local structure of DNA can significantly affect its protein-binding properties (i.e. intrinsic curvature in DNA-histone complexes). In a step towards comprehensive DNA-protein modeling, we expand the 3SPN.2 coarse-grained model to include intrinsic shape, and validate the refined model against experimental data including melting temperature, local flexibility, persistence length, and minor groove width profile. △ Less

Submitted 30 April, 2014; originally announced April 2014.

Comments: 9 pages, 6 figures

Journal ref: J. Chem. Phys. 141,165103 (2014)