File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: Size-independent neural networks based first-principles method for accurate prediction of heat of formation of fuels

TitleSize-independent neural networks based first-principles method for accurate prediction of heat of formation of fuels
Authors
Issue Date2018
PublisherAcademic Press. The Journal's web site is located at http://scitation.aip.org/content/aip/journal/jcp
Citation
The Journal of Chemical Physics, 2018, v. 148 n. 24, p. article no. 241738 How to Cite?
AbstractNeural network-based first-principles method for predicting heat of formation (HOF) was previously demonstrated to be able to achieve chemical accuracy in a broad spectrum of target molecules [L. H. Hu et al., J. Chem. Phys. 119, 11501 (2003)]. However, its accuracy deteriorates with the increase in molecular size. A closer inspection reveals a systematic correlation between the prediction error and the molecular size, which appears correctable by further statistical analysis, calling for a more sophisticated machine learning algorithm. Despite the apparent difference between simple and complex molecules, all the essential physical information is already present in a carefully selected set of small molecule representatives. A model that can capture the fundamental physics would be able to predict large and complex molecules from information extracted only from a small molecules database. To this end, a size-independent, multi-step multi-variable linear regression-neural network-B3LYP method is developed in this work, which successfully improves the overall prediction accuracy by training with smaller molecules only. And in particular, the calculation errors for larger molecules are drastically reduced to the same magnitudes as those of the smaller molecules. Specifically, the method is based on a 164-molecule database that consists of molecules made of hydrogen and carbon elements. 4 molecular descriptors were selected to encode molecule's characteristics, among which raw HOF calculated from B3LYP and the molecular size are also included. Upon the size-independent machine learning correction, the mean absolute deviation (MAD) of the B3LYP/6-311+G(3df,2p)-calculated HOF is reduced from 16.58 to 1.43 kcal/mol and from 17.33 to 1.69 kcal/mol for the training and testing sets (small molecules), respectively. Furthermore, the MAD of the testing set (large molecules) is reduced from 28.75 to 1.67 kcal/mol. © 2018 Author(s).
Persistent Identifierhttp://hdl.handle.net/10722/279297
ISSN
2019 Impact Factor: 2.888
2015 SCImago Journal Rankings: 1.095
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorYang, G-
dc.contributor.authorWu, J-
dc.contributor.authorChen, S-
dc.contributor.authorZhou, W-
dc.contributor.authorSun, J-
dc.contributor.authorChen, G-
dc.date.accessioned2019-10-25T13:53:00Z-
dc.date.available2019-10-25T13:53:00Z-
dc.date.issued2018-
dc.identifier.citationThe Journal of Chemical Physics, 2018, v. 148 n. 24, p. article no. 241738-
dc.identifier.issn0021-9614-
dc.identifier.urihttp://hdl.handle.net/10722/279297-
dc.description.abstractNeural network-based first-principles method for predicting heat of formation (HOF) was previously demonstrated to be able to achieve chemical accuracy in a broad spectrum of target molecules [L. H. Hu et al., J. Chem. Phys. 119, 11501 (2003)]. However, its accuracy deteriorates with the increase in molecular size. A closer inspection reveals a systematic correlation between the prediction error and the molecular size, which appears correctable by further statistical analysis, calling for a more sophisticated machine learning algorithm. Despite the apparent difference between simple and complex molecules, all the essential physical information is already present in a carefully selected set of small molecule representatives. A model that can capture the fundamental physics would be able to predict large and complex molecules from information extracted only from a small molecules database. To this end, a size-independent, multi-step multi-variable linear regression-neural network-B3LYP method is developed in this work, which successfully improves the overall prediction accuracy by training with smaller molecules only. And in particular, the calculation errors for larger molecules are drastically reduced to the same magnitudes as those of the smaller molecules. Specifically, the method is based on a 164-molecule database that consists of molecules made of hydrogen and carbon elements. 4 molecular descriptors were selected to encode molecule's characteristics, among which raw HOF calculated from B3LYP and the molecular size are also included. Upon the size-independent machine learning correction, the mean absolute deviation (MAD) of the B3LYP/6-311+G(3df,2p)-calculated HOF is reduced from 16.58 to 1.43 kcal/mol and from 17.33 to 1.69 kcal/mol for the training and testing sets (small molecules), respectively. Furthermore, the MAD of the testing set (large molecules) is reduced from 28.75 to 1.67 kcal/mol. © 2018 Author(s).-
dc.languageeng-
dc.publisherAcademic Press. The Journal's web site is located at http://scitation.aip.org/content/aip/journal/jcp-
dc.relation.ispartofThe Journal of Chemical Physics-
dc.titleSize-independent neural networks based first-principles method for accurate prediction of heat of formation of fuels-
dc.typeArticle-
dc.identifier.emailWu, J: wu2324@hku.hk-
dc.identifier.emailChen, S: h0992048@hku.hk-
dc.identifier.emailChen, G: ghchen@hku.hk-
dc.identifier.authorityChen, G=rp00671-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1063/1.5024442-
dc.identifier.scopuseid_2-s2.0-85048328659-
dc.identifier.hkuros308207-
dc.identifier.volume148-
dc.identifier.issue24-
dc.identifier.spagearticle no. 241738-
dc.identifier.epagearticle no. 241738-
dc.identifier.isiWOS:000437190300041-
dc.publisher.placeUnited Kingdom-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats