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

Machine learning (ML) is emerging as an essential tool in the molecular modeling community. The most mature applications are in batch learning with data generated from molecular modeling calculations. For example, one can run a molecular dynamics (MD) or density functional theory (DFT) simulation to completion, from which the trajectory is then used as input data to train a deep neural network that reproduces the energetics at a fraction of the computational cost. Some recent examples include an energy-conserving force-field learned with a custom gradient-domain model (Chmiela et al., 2017), DFT-based neural network force-fields (J. S. Smith et al., 2017a), and a neural network coarse-grained potential (Wang et al., 2019). Other applications include the use of ML methods for collective variable (CV) calculation (Trapl et al., 2019) and enhanced sampling (Wang, Lamim Ribeiro, & Tiwary, 2020) for MD simulations. A limitation of these methods is that they treat the molecular modeling calculations as a static dataset, whereas molecular modeling calculations can be treated as interrogative functions, opening the door to methods like reinforcement learning or active learning. Thus there is a clear limitation of existing implementations due to their sequential nature of going from calculation to ML. Another practical issue is that neural network force-field implementations often duplicate standard ML frameworks, preventing them from keeping progress with state-of-the art methods. For example Rupp, Tkatchenko, Müller, & Anatole Von Lilienfeld (2012) and Botu & Ramprasad (2015) are benchmark works in this field and custom implementations. This limits the scope and speed of translating ML advances to the molecular modeling community.

HOOMD-TF: GPU-Accelerated, Online Machine Learning in the HOOMD-blue Molecular Dynamics Engine