GNN and transfer learning for prediction of formation enthalpy of metal hydrides
Апстракт
Prediction of metal hydride formation enthalpy is one of the key elements for a rapid screening and design of new hydrogen storage materials. In the last decades, DFT (density functional theory) approach showed good predictive potential for the ground state properties and calculation of hydride formation energies. Recently, graph neural network (GNN) implementations show promising results for fast and reliable prediction of formation energies for molecules and crystals. Here, we consider approach for universal machine learning based on a MatErials Graph Network (MEGNet) [1] that enable hydride formation energy prediction with a DFT accuracy. We demonstrate wide screening of potential dopants in Mg2FeH6 and Mg2NiH4. In addition, we study the potential of transfer learning for building the universal machine- learning model capable of addressing experimentally reported hydride formation enthalpies.
Извор:
Solid-State Science & Research 2021 : Book of Abstracts and Program, 2021, 67-67Издавач:
- Zagreb : "Ruđer Bošković" Institute
Напомена:
- Solid-State Science & Research ; 10-11th June 2021, Online
Колекције
Институција/група
VinčaTY - CONF AU - Batalović, Katarina AU - Radaković, Jana AU - Paskaš Mamula, Bojana PY - 2021 UR - https://vinar.vin.bg.ac.rs/handle/123456789/11244 AB - Prediction of metal hydride formation enthalpy is one of the key elements for a rapid screening and design of new hydrogen storage materials. In the last decades, DFT (density functional theory) approach showed good predictive potential for the ground state properties and calculation of hydride formation energies. Recently, graph neural network (GNN) implementations show promising results for fast and reliable prediction of formation energies for molecules and crystals. Here, we consider approach for universal machine learning based on a MatErials Graph Network (MEGNet) [1] that enable hydride formation energy prediction with a DFT accuracy. We demonstrate wide screening of potential dopants in Mg2FeH6 and Mg2NiH4. In addition, we study the potential of transfer learning for building the universal machine- learning model capable of addressing experimentally reported hydride formation enthalpies. PB - Zagreb : "Ruđer Bošković" Institute C3 - Solid-State Science & Research 2021 : Book of Abstracts and Program T1 - GNN and transfer learning for prediction of formation enthalpy of metal hydrides SP - 67 EP - 67 UR - https://hdl.handle.net/21.15107/rcub_vinar_11244 ER -
@conference{ author = "Batalović, Katarina and Radaković, Jana and Paskaš Mamula, Bojana", year = "2021", abstract = "Prediction of metal hydride formation enthalpy is one of the key elements for a rapid screening and design of new hydrogen storage materials. In the last decades, DFT (density functional theory) approach showed good predictive potential for the ground state properties and calculation of hydride formation energies. Recently, graph neural network (GNN) implementations show promising results for fast and reliable prediction of formation energies for molecules and crystals. Here, we consider approach for universal machine learning based on a MatErials Graph Network (MEGNet) [1] that enable hydride formation energy prediction with a DFT accuracy. We demonstrate wide screening of potential dopants in Mg2FeH6 and Mg2NiH4. In addition, we study the potential of transfer learning for building the universal machine- learning model capable of addressing experimentally reported hydride formation enthalpies.", publisher = "Zagreb : "Ruđer Bošković" Institute", journal = "Solid-State Science & Research 2021 : Book of Abstracts and Program", title = "GNN and transfer learning for prediction of formation enthalpy of metal hydrides", pages = "67-67", url = "https://hdl.handle.net/21.15107/rcub_vinar_11244" }
Batalović, K., Radaković, J.,& Paskaš Mamula, B.. (2021). GNN and transfer learning for prediction of formation enthalpy of metal hydrides. in Solid-State Science & Research 2021 : Book of Abstracts and Program Zagreb : "Ruđer Bošković" Institute., 67-67. https://hdl.handle.net/21.15107/rcub_vinar_11244
Batalović K, Radaković J, Paskaš Mamula B. GNN and transfer learning for prediction of formation enthalpy of metal hydrides. in Solid-State Science & Research 2021 : Book of Abstracts and Program. 2021;:67-67. https://hdl.handle.net/21.15107/rcub_vinar_11244 .
Batalović, Katarina, Radaković, Jana, Paskaš Mamula, Bojana, "GNN and transfer learning for prediction of formation enthalpy of metal hydrides" in Solid-State Science & Research 2021 : Book of Abstracts and Program (2021):67-67, https://hdl.handle.net/21.15107/rcub_vinar_11244 .