English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Molecular dynamics simulations are powerful tools to extract the microscopicmechanisms characterizing the properties of soft materials. We recentlyintroduced machine learning surrogates for molecular dynamics simulations ofsoft materials and demonstrated that artificial neural network based regressionmodels can successfully predict the relationships between the input materialattributes and the simulation outputs. Here, we show that statisticaluncertainties associated with the outputs of molecular dynamics simulations canbe utilized to train artificial neural networks and design machine learningsurrogates with higher accuracy and generalizability. We design soft labels forthe simulation outputs by incorporating the uncertainties in the estimatedaverage output quantities, and introduce a modified loss function thatleverages these soft labels during training to significantly reduce thesurrogate prediction error for input systems in the unseen test data. Theapproach is illustrated with the design of a surrogate for molecular dynamicssimulations of confined electrolytes to predict the complex relationshipbetween the input electrolyte attributes and the output ionic structure. Thesurrogate predictions for the ionic density profiles show excellent agreementwith the ground truth results produced using molecular dynamics simulations.The high accuracy and small inference times associated with the surrogatepredictions provide quick access to quantities derived using the number densityprofiles and facilitate rapid sensitivity analysis.

Designing Machine Learning Surrogates using Outputs of Molecular Dynamics Simulations as Soft Labels