Graph Neural Network(GNN)potentials relying on chemical locality offer near-quantum mechanical accuracy at significantly reduced computational costs.Message-passing GNNs model interactions beyond their immediate neigh...Graph Neural Network(GNN)potentials relying on chemical locality offer near-quantum mechanical accuracy at significantly reduced computational costs.Message-passing GNNs model interactions beyond their immediate neighborhood by propagating local information between neighboring particles while remaining effectively local.However,locality precludes modeling long-range effects critical to many real-world systems,such as charge transfer,electrostatic interactions,and dispersion effects.In this work,we propose the Charge Equilibration Layer for Long-range Interactions(CELLI)to address the challenge of efficiently modeling non-local interactions.This novel architecture generalizes the classical charge equilibration(Qeq)method to a model-agnostic building block for modern equivariant GNN potentials.Therefore,CELLI extends the capability of GNNs to model longrange interactions while providing high interpretability through explicitly modeled charges.On benchmark systems,CELLI achieves state-of-the-art results for strictly local models.CELLI generalizes to diverse datasets and large structureswhile providing high computational efficiency and robust predictions.展开更多
In this paper,we propose a novel flexible optimization pipeline for determining the optimal adsorption sites,named AUGUR(Aware of Uncertainty Graph Unit Regression).Our model combines graph neural networks and Gaussia...In this paper,we propose a novel flexible optimization pipeline for determining the optimal adsorption sites,named AUGUR(Aware of Uncertainty Graph Unit Regression).Our model combines graph neural networks and Gaussian processes to create a flexible,efficient,symmetry-aware,translation,and rotation-invariant predictor with inbuilt uncertainty quantification.This predictor is then used as a surrogate for a data-efficient Bayesian Optimization scheme to determine the optimal adsorption positions.This pipeline determines the optimal position of large and complicated clusters with far fewer iterations than current state-of-the-art approaches.Further,it does not rely on hand-crafted features and can be seamlessly employed on any molecule without any alterations.Additionally,the pooling properties of graphs allow for the processing of molecules of different sizes by the same model.This allows the energy prediction ofcomputationally demanding systemsby a model trained on comparatively smaller and less expensive ones.展开更多
Infrared(IR)spectroscopy is a pivotal analytical tool as it provides real-time molecular insight into material structures and enables the observation of reaction intermediates in situ.However,interpreting IR spectra o...Infrared(IR)spectroscopy is a pivotal analytical tool as it provides real-time molecular insight into material structures and enables the observation of reaction intermediates in situ.However,interpreting IR spectra often requires high-fidelity simulations,such as density functional theory based ab-initio molecular dynamics,which are computationally expensive and therefore limited in the tractable system size and complexity.In this work,we present a novel active learning-based framework,implemented in the open-source software package PALIRS,for efficiently predicting the IR spectra of small catalytically relevant organic molecules.PALIRS leverages active learning to train a machine-learned interatomic potential,which is then used for machine learning-assisted molecular dynamics simulations to calculate IR spectra.PALIRS reproduces IR spectra computed with ab-initio molecular dynamics accurately at a fraction of the computational cost.PALIRS further agrees well with available experimental data not only for IR peak positions but also for their amplitudes.This advancement with PALIRS enables high-throughput prediction of IR spectra,facilitating the exploration of larger and more intricate catalytic systems and aiding the identification of novel reaction pathways.展开更多
Transforming CO_(2)into methanol represents a crucial step towards closing the carbon cycle,with thermoreduction technology nearing industrial application.However,obtaining high methanol yields and ensuring the stabil...Transforming CO_(2)into methanol represents a crucial step towards closing the carbon cycle,with thermoreduction technology nearing industrial application.However,obtaining high methanol yields and ensuring the stability of heterocatalysts remain significant challenges.Herein,we present a sophisticated computational framework to accelerate the discovery of thermal heterogeneous catalysts,using machine-learned force fields.We propose a new catalytic descriptor,termed adsorption energy distribution,that aggregates the binding energies for different catalyst facets,binding sites,and adsorbates.The descriptor is versatile and can be adjusted to a specific reaction through careful choice of the key-step reactants and reaction intermediates.By applying unsupervised machine learning and statistical analysis to a dataset comprising nearly 160 metallic alloys,we offer a powerful tool for catalyst discovery.We propose new promising candidates such as ZnRh and ZnPt_(3),which to our knowledge,have not yet been tested,and discuss their possible advantage in terms of stability.展开更多
Imaging offers a fast and accessible means for spatial characterization of halide perovskite photovoltaic materials,yet extracting optoelectrical properties—such as power conversion efficiency(PCE)—remains challengi...Imaging offers a fast and accessible means for spatial characterization of halide perovskite photovoltaic materials,yet extracting optoelectrical properties—such as power conversion efficiency(PCE)—remains challenging.This study presents a deep learning methodology that correlates optical reflective images of perovskite solar cells with their PCE by focusing on image differences rather than absolute visual features.The approach predicts relative changes in PCE by comparing images of the same device in different states(e.g.,before and after encapsulation)or against a reference image.This comparative technique significantly outperforms traditional methods that attempt to directly infer PCE from a single image.Furthermore,it demonstrates high effectiveness in low-data regimes,using only 115 samples.By leveraging convolutional neural networks(CNNs)trained on small datasets,the method offers an adaptable and scalable solution for device characterization.Overall,the comparative approach enhances the accuracy and applicability of machine vision in perovskite solar cell analysis.展开更多
基金Funded by the European Union. Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council Executive AgencyNeither the European Union nor the granting authority can be held responsible for them. This work was funded by the ERC (StG SupraModel) - 101077842the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - 534045056 and 561190767.
文摘Graph Neural Network(GNN)potentials relying on chemical locality offer near-quantum mechanical accuracy at significantly reduced computational costs.Message-passing GNNs model interactions beyond their immediate neighborhood by propagating local information between neighboring particles while remaining effectively local.However,locality precludes modeling long-range effects critical to many real-world systems,such as charge transfer,electrostatic interactions,and dispersion effects.In this work,we propose the Charge Equilibration Layer for Long-range Interactions(CELLI)to address the challenge of efficiently modeling non-local interactions.This novel architecture generalizes the classical charge equilibration(Qeq)method to a model-agnostic building block for modern equivariant GNN potentials.Therefore,CELLI extends the capability of GNNs to model longrange interactions while providing high interpretability through explicitly modeled charges.On benchmark systems,CELLI achieves state-of-the-art results for strictly local models.CELLI generalizes to diverse datasets and large structureswhile providing high computational efficiency and robust predictions.
基金funding from the project ProperPhotoMile,supported under the umbrella of SOLAR-ERA.NET Cofund 2 by The Spanish Ministry of Science and Education and the AEI under the project PCI2020-112185 and CDTI project number IDI-20210171the Federal Ministry for Economic Affairs and Energy on the basis of a decision by the German Bundestag project number FKZ 03EE1070B and FKZ 03EE1070A+2 种基金the Israel Ministry of Energy with project number 220-11-031.SOLAR-ERA.NET is supported by the European Commission within the EU Framework Program for Research and Innovation HORIZON 2020(Cofund ERA-NET Action,786483)Further,A.G.acknowledges financial support from TUM Innovation Network for Artificial Intelligence powered Multifunctional Material Design(ARTEMIS)funding in the framework of Deutsche Forschungsgemeinschaft(DFG,German Research Foundation)under Germany’s Excellence Strategy-EXC 2089/1-390776260(e-conversion).Lastly,we wish to express our gratitude to Dr.Inigo Iribarren for creating the flowcharts for this work.
文摘In this paper,we propose a novel flexible optimization pipeline for determining the optimal adsorption sites,named AUGUR(Aware of Uncertainty Graph Unit Regression).Our model combines graph neural networks and Gaussian processes to create a flexible,efficient,symmetry-aware,translation,and rotation-invariant predictor with inbuilt uncertainty quantification.This predictor is then used as a surrogate for a data-efficient Bayesian Optimization scheme to determine the optimal adsorption positions.This pipeline determines the optimal position of large and complicated clusters with far fewer iterations than current state-of-the-art approaches.Further,it does not rely on hand-crafted features and can be seamlessly employed on any molecule without any alterations.Additionally,the pooling properties of graphs allow for the processing of molecules of different sizes by the same model.This allows the energy prediction ofcomputationally demanding systemsby a model trained on comparatively smaller and less expensive ones.
基金funding from Horizon Europe MSCA Doctoral network grant n.101073486, EUSpecLab, funded by the European UnionO.K. and P.R. have received funding from the European Union – NextGenerationEU instrument and are funded by the Research Council of Finland (grant numbers 348179, 346377, and 364227). We acknowledge CSC, Finland for awarding access to the LUMI supercomputer, owned by the EuroHPC Joint Undertaking, hosted by CSC (Finland) and the LUMI consortium through CSC, Finland, extreme-scale project ALVS. The authors also gratefully acknowledge the additional computational resources provided by CSC – IT Center for Science, Finland, and the Aalto Science-IT project.
文摘Infrared(IR)spectroscopy is a pivotal analytical tool as it provides real-time molecular insight into material structures and enables the observation of reaction intermediates in situ.However,interpreting IR spectra often requires high-fidelity simulations,such as density functional theory based ab-initio molecular dynamics,which are computationally expensive and therefore limited in the tractable system size and complexity.In this work,we present a novel active learning-based framework,implemented in the open-source software package PALIRS,for efficiently predicting the IR spectra of small catalytically relevant organic molecules.PALIRS leverages active learning to train a machine-learned interatomic potential,which is then used for machine learning-assisted molecular dynamics simulations to calculate IR spectra.PALIRS reproduces IR spectra computed with ab-initio molecular dynamics accurately at a fraction of the computational cost.PALIRS further agrees well with available experimental data not only for IR peak positions but also for their amplitudes.This advancement with PALIRS enables high-throughput prediction of IR spectra,facilitating the exploration of larger and more intricate catalytic systems and aiding the identification of novel reaction pathways.
基金funding from the European Union – NextGenerationEU instrument and the Research Council of Finland's AICon project (grant number no. 348179). The authors gratefully acknowledge CSC – IT Center for Science, Finland, and the Aalto Science-IT project for generous computational resources.
文摘Transforming CO_(2)into methanol represents a crucial step towards closing the carbon cycle,with thermoreduction technology nearing industrial application.However,obtaining high methanol yields and ensuring the stability of heterocatalysts remain significant challenges.Herein,we present a sophisticated computational framework to accelerate the discovery of thermal heterogeneous catalysts,using machine-learned force fields.We propose a new catalytic descriptor,termed adsorption energy distribution,that aggregates the binding energies for different catalyst facets,binding sites,and adsorbates.The descriptor is versatile and can be adjusted to a specific reaction through careful choice of the key-step reactants and reaction intermediates.By applying unsupervised machine learning and statistical analysis to a dataset comprising nearly 160 metallic alloys,we offer a powerful tool for catalyst discovery.We propose new promising candidates such as ZnRh and ZnPt_(3),which to our knowledge,have not yet been tested,and discuss their possible advantage in terms of stability.
基金EXC 2089:e-conversion DFG-cluster of excellence-TUM innovation network,Technical University of Munich,funded through the German Excellence Initiative and the state of Bavaria(A.G.,M.H.).Support by the project ProperPhotoMile is also gratefully acknowledged(A.G.,I.V-F.),under the umbrella of SOLAR-ERA.NET Cofund 2 by The Spanish Ministry of Science and Education and the AEI under the project PCI2020-112185 and CDTI project number IDI-20210171the Federal Ministry for Economic Affairs and Energy on the basis of a decision by the German Bundestag project number FKZ03EE1070B and FKZ 03EE1070A+2 种基金and the Israel Ministry of Energy with project number 220-11-031SOLAR-ERA.NET is supportedby the European Commission within the EU Framework Programme for Research and Innovation HORIZON 2020(Cofund ERA-NETAction,N 786483)D.K.K.is grateful for the Blaustein postdoctoral fellowship at BGU.R.K.G.is grateful for the Swiss Inst.of Dryland Environmental and Energy Research postdoctoral fellowship at BGU.The authors are also grateful for partial support by the Israel Ministry of Energy,project number 222-11-081.
文摘Imaging offers a fast and accessible means for spatial characterization of halide perovskite photovoltaic materials,yet extracting optoelectrical properties—such as power conversion efficiency(PCE)—remains challenging.This study presents a deep learning methodology that correlates optical reflective images of perovskite solar cells with their PCE by focusing on image differences rather than absolute visual features.The approach predicts relative changes in PCE by comparing images of the same device in different states(e.g.,before and after encapsulation)or against a reference image.This comparative technique significantly outperforms traditional methods that attempt to directly infer PCE from a single image.Furthermore,it demonstrates high effectiveness in low-data regimes,using only 115 samples.By leveraging convolutional neural networks(CNNs)trained on small datasets,the method offers an adaptable and scalable solution for device characterization.Overall,the comparative approach enhances the accuracy and applicability of machine vision in perovskite solar cell analysis.
