Equation of State of Fluid Methane from First Principles with Machine Learning Potentials

Veit, Max; Jain, Sandeep Kumar; Bonakala, Satyanarayana; Rudra, Indranil; Hohl, Detlef; Csänyi, Gábor

doi:10.1021/acs.jctc.8b01242

Cited by 68 publications

(90 citation statements)

References 138 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use the Gaussian Approximation Potential (GAP) (Bartok et al, 2010), essentially a kernel ridge regression method (Kung, 2014). This is just one of a class of recently popularized machine learning methods for creating nonparametric interatomic potentials, which has been shown to be very successful in tackling difficult materials modelling problems, ranging from investigating the structure of amorphous materials (carbon (Deringer et al, 2017(Deringer et al, , 2019, silicon (Bartók et al, 2018)), the mechanics of metals (tungsten (Szlachta et al, 2014), iron (Dragoni et al, 2018)) to molecular liquids such (water (Bartók et al, 2013a), methane (Veit et al, 2019). There are many alternatives, using other regression frameworks, such as artificial neural networks (Behler and Parrinello, 2007) and even linear regression (Shapeev, 2017;Drautz, 2019).…”

Section: Machine Learning Potentialsmentioning

confidence: 99%

Partitioning of sulfur between solid and liquid iron under Earth’s core conditions: Constraints from atomistic simulations with machine learning potentials

Zhang

Csänyi

2020

Geochimica et Cosmochimica Acta

Self Cite

View full text Add to dashboard Cite

Partition coefficients of light elements between the solid and liquid iron phases are crucial for uncovering the state and dynamics of the Earth's core. As one of the major light element candidates, sulfur has attracted extensive interests for measuring its partitioning and phase behaviors over the last several decades, but the relevant experimental data under Earth's core conditions are still scarce. In this study, using a toolkit consisting of electronic structure theory, high-accuracy machine learning potentials and rigorous free energy calculations, we establish an efficient and extendible framework for predicting complex phase behaviors of iron alloys under extreme conditions. As a first application of this framework, we predict the partition coefficients of sulfur over wide range of temperatures and pressures (from 4000 K, 150 GPa to 6000 K, 330 GPa), which are demonstrated to be in good agreement with previous experiments and ab initio simulations. After a continuous increase below ~250 GPa, the partition coefficient is found to be around 0.750.07 at higher pressures and are essentially temperature-independent. Given these predictions, the partitioning of sulfur is confirmed to be insufficient to account for the observed density jump across the Earth's inner core boundary and its roles on the geodynamics of the Earth's core should be minor.

show abstract

Section: Machine Learning Potentialsmentioning

confidence: 99%

Partitioning of sulfur between solid and liquid iron under Earth’s core conditions: Constraints from atomistic simulations with machine learning potentials

Zhang

Csänyi

2020

Geochimica et Cosmochimica Acta

Self Cite

View full text Add to dashboard Cite

show abstract

“…As with the water models above, to quantitatively evaluate bulk properties training an accurate reference method and the inclusion of nuclear quantum effects would be necessary. 33,35 Nevertheless, this example already demonstrates that our training method is applicable to a range of chemical systems beyond water or aqueous solutions. Table 1.…”

Section: Other Solvent Systemsmentioning

confidence: 92%

“…26 In the present work, we employ the Gaussian Approximation Potential (GAP) framework, 21 which has been used to generate force fields for a range of elemental, [27][28][29] multicomponent inorganic, 30,31 and recently gas-phase organic systems. 14,32 Initial studies of condensed phase molecular systems with GAPs include fluid methane 33 and phosphorus. 34 These potentials, while accurate, have required considerable computational effort and human oversight.…”

Section: Introductionmentioning

confidence: 99%

A Transferable Active-Learning Strategy for Reactive Molecular Force Fields

Young¹,

Johnston-Wood²,

Deringer³

et al. 2021

Preprint

View full text Add to dashboard Cite

<p>Predictive simulations of dynamic processes in molecular systems require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct force-field models for large-scale molecular simulation by fitting to high-level quantum-mechanical data. However, machine-learned force fields generally require considerable human intervention and data volume. Here we show that, by leveraging hierarchical and active learning, accurate Gaussian Approximation Potential (GAP) models for diverse chemical systems can be developed in an autonomous way, requiring only hundreds to a few thousand energy and gradient evaluations on the reference potential-energy surface. Our approach relies on a decomposition of the condensed-phase molecular system into intra- and inter-molecular terms, and on the definition of a prospective error metric to quantify accuracy. We demonstrate applications to a range of molecular systems: from bulk water, organic solvents, and a solvated ion onwards to the description of chemical reactivity, including, a bifurcating Diels–Alder reaction in the gas phase and non-equilibrium dynamics (S<sub>N</sub>2 reaction) in explicit solvent. The method provides a route to routinely generating machine-learned force fields for complex and/or reactive molecular systems. </p>

show abstract

“… 76 For example, physics-guided breakdown of the target proved to be useful in the creation of a model for the equation of state of fluid methane. 77 …”

Section: Machine Learning Landscapementioning

confidence: 99%

Big-Data Science in Porous Materials: Materials Genomics and Machine Learning

et al. 2020

View full text Add to dashboard Cite

By combining metal nodes with organic linkers we can potentially synthesize millions of possible metal–organic frameworks (MOFs). The fact that we have so many materials opens many exciting avenues but also create new challenges. We simply have too many materials to be processed using conventional, brute force, methods. In this review, we show that having so many materials allows us to use big-data methods as a powerful technique to study these materials and to discover complex correlations. The first part of the review gives an introduction to the principles of big-data science. We show how to select appropriate training sets, survey approaches that are used to represent these materials in feature space, and review different learning architectures, as well as evaluation and interpretation strategies. In the second part, we review how the different approaches of machine learning have been applied to porous materials. In particular, we discuss applications in the field of gas storage and separation, the stability of these materials, their electronic properties, and their synthesis. Given the increasing interest of the scientific community in machine learning, we expect this list to rapidly expand in the coming years.

show abstract

Equation of State of Fluid Methane from First Principles with Machine Learning Potentials

Cited by 68 publications

References 138 publications

Partitioning of sulfur between solid and liquid iron under Earth’s core conditions: Constraints from atomistic simulations with machine learning potentials

Partitioning of sulfur between solid and liquid iron under Earth’s core conditions: Constraints from atomistic simulations with machine learning potentials

A Transferable Active-Learning Strategy for Reactive Molecular Force Fields

Big-Data Science in Porous Materials: Materials Genomics and Machine Learning

Contact Info

Product

Resources

About