Predicting molecular properties is essential for advancing for advancing drug discovery and design. Recently, Graph Neural Networks (GNNs) have gained prominence due to their ability to capture the complex structural ...Predicting molecular properties is essential for advancing for advancing drug discovery and design. Recently, Graph Neural Networks (GNNs) have gained prominence due to their ability to capture the complex structural and relational information inherent in molecular graphs. Despite their effectiveness, the “black-box” nature of GNNs remains a significant obstacle to their widespread adoption in chemistry, as it hinders interpretability and trust. In this context, several explanation methods based on factual reasoning have emerged. These methods aim to interpret the predictions made by GNNs by analyzing the key features contributing to the prediction. However, these approaches fail to answer critical questions: “How to ensure that the structure-property mapping learned by GNNs is consistent with established domain knowledge”. In this paper, we propose MMGCF, a novel counterfactual explanation framework designed specifically for the prediction of GNN-based molecular properties. MMGCF constructs a hierarchical tree structure on molecular motifs, enabling the systematic generation of counterfactuals through motif perturbations. This framework identifies causally significant motifs and elucidates their impact on model predictions, offering insights into the relationship between structural modifications and predicted properties. Our method demonstrates its effectiveness through comprehensive quantitative and qualitative evaluations of four real-world molecular datasets.展开更多
In order to analyze the origin of carbon monoxide(CO) in coal seams, stress–strain experiments under temperature of 50, 150 and 250 °C were conducted using lignite from Kailuan mining area. Fourier transform inf...In order to analyze the origin of carbon monoxide(CO) in coal seams, stress–strain experiments under temperature of 50, 150 and 250 °C were conducted using lignite from Kailuan mining area. Fourier transform infrared spectroscopy and elemental analysis were carried out before and after deformation of the samples. The results indicated that CO generated at 150 and 250 °C; the gas component was mostly oxygen(O_2), with small amount of carbon dioxide(CO_2), methane(CH_4) and hydrogen(H_2). At 50 °C, O_2 and a little CO_2 were observed and no CO was found. The carbon content of the coal samples increased slightly after deformation, and the oxygen content, H/C ratio, and O/C ratio decreased. The molecular structure of coal displayed different evolution characteristics at various temperatures. At 50 and 150 °C, the falling off of side chains, broken of ether bond and directional realignment of the aliphatic chains resulting in the formation of long chains were the main performance of coal molecular structure evolution. While at 250 °C, the side chains fell off and short chains formed. Furthermore, at both 150 and 250 °C, condensed degree of aromatic ring increased. Under the action of temperature and pressure, CO forms in two ways.The first is that ether bond breaks, oxygen and carbon atoms combine together and forms CO, or O_2 forming in the broken of ether–oxygen bond leads to the oxidation of free radicals and resulting in the formation of CO. And the second is that CO derives from falling off of C=O group.展开更多
Using Bohmlan trajectory （151） method, we investigate the dynamic interference in nlgn-orcter narmonlc generauon from diatomic molecular ions. It is demonstrated that the main characteristics of the molecular harmon...Using Bohmlan trajectory （151） method, we investigate the dynamic interference in nlgn-orcter narmonlc generauon from diatomic molecular ions. It is demonstrated that the main characteristics of the molecular harmonic spectrum can be well reproduced by only two BTs which are located at the two ions. This haiapens because these two localized trajectories can receive and store the whole collision information coming from all of the other re-collision trajectories. Therefore, the amplitudes and frequencies of these two trajectories represent the intensity and frequency distribution of the harmonic generation. Moreover, the interference between these two trajectories shows a dip in the harmonic spectrum, which reveals the molecular structure information.展开更多
We propose an efficient method for the generation of an isolated attosecond pulse from the asymmetric molecular ions HeH^2＋ by adding a half-cycle-like field （HCLF） to the fundamental driving laser field. The high-...We propose an efficient method for the generation of an isolated attosecond pulse from the asymmetric molecular ions HeH^2＋ by adding a half-cycle-like field （HCLF） to the fundamental driving laser field. The high-order harmonic generation （HHG） is investigated by numerically sowing the time-dependent Schrodinger equation. By performing the time-frequency distributions and the electronic wave packet probability densities, we find that the optimizing combined field is not only useful for extending the HHG cutoff, but also for simplifying the recombination channels through controlling the electron localization. In addition, by adjusting the intensity of the HCLF, a dominant short quantum path is selected to contribute the HHG spectrum. As a result, a 75-as isolated attosecond pulse is obtained by superposing a proper range of the harmonics.展开更多
Polyimides are widely used as chip passivation layers and organic substrates in microelectronic packaging. Plasma treatment has been used to enhance the interfacial properties of polyimides, but its molecular mechanis...Polyimides are widely used as chip passivation layers and organic substrates in microelectronic packaging. Plasma treatment has been used to enhance the interfacial properties of polyimides, but its molecular mechanism is not clear. In this research, the effects of polyimide surface plasma treatment on the molecular structures at corresponding polyimide/air and buried polyimideJepoxy interfaces were investigated in situ using sum frequency generation （SFG） vibrational spectroscopy. SFG results show that the polyimide backbone molecular structure was different at polyimide/air and polyimideJepoxy interfaces before and after plasma treatment. The different molecular structures at each interface indicate that structural reordering of the polyimide backbone occurred as a result of plasma treatment and contact with the epoxy adhesive. Furthermore, quantitative orientation analysis indicated that plasma treatment of polyimide surfaces altered the twist angle of the polyimide backbone at corresponding buried polyimide/epoxy interfaces. These SFG results indicate that plasma treatment of polymer surfaces can alter the molecular structure at corresponding polymer/air and buried polymer interfaces.展开更多
Permanent dipole moments induced high-order harmonic generation(HHG)signals offer a potential approach to producing elliptically or even circularly polarized X-ray attosecond sources.Previous studies on this topic hav...Permanent dipole moments induced high-order harmonic generation(HHG)signals offer a potential approach to producing elliptically or even circularly polarized X-ray attosecond sources.Previous studies on this topic have mainly focused on diatomic molecules such as CO and HeH.Based on this scheme,significant HHG signals in the direction perpendicular to the molecular axis can be observed in both the high-energy and low-energy regions.However,we found that the high-order harmonics induced by the permanent dipole moments of polyatomic complex molecules involve more intricate physical processes.Using time-dependent density functional theory,we simulated the dynamics of HHG from NH2COOH and NH2COSH interacting with linearly polarized lasers.We found that the harmonic signals in the direction perpendicular to the N-C bond were significantly enhanced in the high-energy photon region.Our analysis indicates that this is due to the complex molecular configuration of NH_(2)COOH and NH_(2)COSH:while the NH_(2) group has C_(2v) symmetry,both COOH and COSH groups lack this symmetry.This structural characteristic results in permanent dipole moments being felt only when electrons return to either COSH or COOH groups,but not to NH_(2) group.Additionally,our results reveal a multi-plateau structure in HHG signal along laser polarization direction,a phenomenon arising from multi-electron and multiorbital effects during interaction between complex molecule and strong laser field.展开更多
Current biodegradation timelines show that polyesters take over 200 years to break down. A crucial component of several industries, polyesters are relied upon for materials development and thus require sustainable alt...Current biodegradation timelines show that polyesters take over 200 years to break down. A crucial component of several industries, polyesters are relied upon for materials development and thus require sustainable alternatives. Recent works in generative modeling have made it possible to produce large sets of chemical structures, but current molecular screening methods are expensive, not scalable, and are oversimplified. This work evaluates whether a molecule’s biodegradability potential can be accurately predicted by training a model on recent experimental data. Additionally, three chemical descriptors were evaluated on the final molecules for their effects on biodegradability: molecular structure, bond types, and solubility. A Gradient Boosted Machine was trained on a dataset of 600 molecules and their binary labels on biodegradability. The classification model effectively captured the biodegradability property, yielding an Area Under the Receiver Operating Characteristics, AUROC, of 84% and an Area Under the Precision Recall Curve, or AUPRC, of 87%. Additionally, an existing amortized synthetic tree generation model, SynNet, validated each molecule by showing chemical synthesizability and producing simple and interpretable synthesis pathways. This approach of filtering by prediction and chemical rule interpretation is inexpensive, highly scalable and can capture the necessary complexity. Using this method, novel polyester candidates can be polymerized and produced into sustainable fabrics, reducing environmental stress from textile-reliant industries.展开更多
Molecular graph generation is an emerging area of research withnumerous applications. Most existing approaches generate new molecules bylearning a generative model from a collection of known molecules. However,how to ...Molecular graph generation is an emerging area of research withnumerous applications. Most existing approaches generate new molecules bylearning a generative model from a collection of known molecules. However,how to obtain the molecular structure from its atomic composition is a problemto be solved. In this paper, it proposes an autoencoder-based model called A2Strwhich takes the atomic composition of a molecular as input. Its molecular structurewas generated and model structures were designed. The effects were verified oneby one. In addition, experiments were implemented with isomer generation anddetermined that the ratio of more than 3% of molecules with different structurescan be obtained from the same combination of atoms.展开更多
The laser phase effect on the spatial distribution of the molecular high-order harmonic generation（MHHG） spectrum from H_2~＋ is theoretically investigated through solving the Non-Bohn-Oppenheimer（NBO） time-depend...The laser phase effect on the spatial distribution of the molecular high-order harmonic generation（MHHG） spectrum from H_2~＋ is theoretically investigated through solving the Non-Bohn-Oppenheimer（NBO） time-dependent Schrodinger equation（TDSE）.The results are shown as follows,（i） The generated harmonics from the two nuclei each present an asymmetric distribution.Particularly,when the laser phases are chosen from 0.0π to 0.6π and from 1.7π to 2.0π,the contribution from the negative-H plays a main role in harmonic generation.When the laser phases are chosen from 0.7πto 1.6k,the contribution from the positive-H to the harmonic generation is remarkably enhanced and becomes greater than that from the negative-H.The electron localization,the time-frequency analyses of the harmonic spectrum and the time-dependent wave function are shown to explain the asymmetric harmonic distribution in H_2~＋,which provides us with a method to control the electron motion in molecules,（ⅱ） As the pulse duration increases,the asymmetric distributions of the MHHG in two H nuclei decrease,（ⅲ） Isotope investigation shows that the asymmetric harmonic distribution can be reduced by introducing the heavy nucleus（i.e.,D_2~＋）.展开更多
A new material for both magnetic coupling and electrocatalytic hydrogen generation based on a copper complex,[（HL）CuCl-CuCl（HL）]HCl 1 is prepared by the reaction of 2-（pyridylmethyl）amino-N,N-bis（2-methylene-4,...A new material for both magnetic coupling and electrocatalytic hydrogen generation based on a copper complex,[（HL）CuCl-CuCl（HL）]HCl 1 is prepared by the reaction of 2-（pyridylmethyl）amino-N,N-bis（2-methylene-4,6-difluorophenol）（H2L） and CuCl2·2H2O.In solid,complex 1 is built from two copper units（[（HL）CuCl]）,and exhibits an antiferromagnetic exchange interaction between copper（Ⅱ） ions（J=-160cm^-1）.In liquid,1 can electrocatalyze hydrogen generation both from acetic acid with a turnover frequency（TOF） of 16.3 moles of hydrogen per mole of catalyst per hour at an overpotential（OP）of 941.6 mV（in DMF）,and a neutral buffer with a TOF of 1415.6 moles of hydrogen per mole of catalyst per hour at an OP of 787.6 mV.展开更多
Generating molecules with desired properties is an important task in chemistry and pharmacy.An efficient method may have a positive impact on finding drugs to treat diseases like COVID-19.Data mining and artificial in...Generating molecules with desired properties is an important task in chemistry and pharmacy.An efficient method may have a positive impact on finding drugs to treat diseases like COVID-19.Data mining and artificial intelligence may be good ways to find an efficient method.Recently,both the generative models based on deep learning and the work based on genetic algorithms have made some progress in generating molecules and optimizing the molecule’s properties.However,existing methods have defects in the experimental evaluation standards.These methods also need to be improved in efficiency and performance.To solve these problems,we propose a method named the Chemical Genetic Algorithm for Large Molecular Space(CALM).Specifically,CALM employs a scalable and efficient molecular representation called molecular matrix.And we design corresponding crossover,mutation,and mask operators inspired by domain knowledge and previous studies.We apply our genetic algorithm to several tasks related to molecular property optimization and constraint molecular optimization.The results of these tasks show that our approach outperforms the other state-of-the-art deep learning and genetic algorithm methods,where the z tests performed on the results of several experiments show that our method is more than 99%likely to be significant.At the same time,based on the experimental results,we point out the defects in the experimental evaluation standard which affects the fair evaluation of all previous work.Avoiding these defects helps to objectively evaluate the performance of all work.展开更多
Advances in producing tailored ultrashort laser pulses have enabled the generation and control of molecular dissociative Rydberg excitation along the polarization axis of the laser field.Here,we exploit the orthogonal...Advances in producing tailored ultrashort laser pulses have enabled the generation and control of molecular dissociative Rydberg excitation along the polarization axis of the laser field.Here,we exploit the orthogonally polarized two-color femtosecond laser fields and achieve an unprecedented two-dimensional control of Rydberg fragment emission in the dissociative frustrated single ionization of oxygen.The Rydberg fragments are collected over the 4πsolid angle,whose momentum distribution is manifested in a characteristic four-lobe pattern.Through precise scanning of the relative phase of the orthogonal two-color laser fields,we demonstrate control over asymmetric directional emission of the Rydberg fragments.Our experimental findings are well supported by classical trajectory Monte Carlo simulations,which suggest an efficient emission control achieved through the manipulation of charge localization upon ionization.展开更多
文摘Predicting molecular properties is essential for advancing for advancing drug discovery and design. Recently, Graph Neural Networks (GNNs) have gained prominence due to their ability to capture the complex structural and relational information inherent in molecular graphs. Despite their effectiveness, the “black-box” nature of GNNs remains a significant obstacle to their widespread adoption in chemistry, as it hinders interpretability and trust. In this context, several explanation methods based on factual reasoning have emerged. These methods aim to interpret the predictions made by GNNs by analyzing the key features contributing to the prediction. However, these approaches fail to answer critical questions: “How to ensure that the structure-property mapping learned by GNNs is consistent with established domain knowledge”. In this paper, we propose MMGCF, a novel counterfactual explanation framework designed specifically for the prediction of GNN-based molecular properties. MMGCF constructs a hierarchical tree structure on molecular motifs, enabling the systematic generation of counterfactuals through motif perturbations. This framework identifies causally significant motifs and elucidates their impact on model predictions, offering insights into the relationship between structural modifications and predicted properties. Our method demonstrates its effectiveness through comprehensive quantitative and qualitative evaluations of four real-world molecular datasets.
基金financial support from the National Science foundation of China(No.41430317)the Discipline Innovative Engineering Plan sponsored by the Ministry of Education of China+1 种基金the State Administration of Foreign Experts Affairs of China(No.13023)the Priority Academic Program Development of Jiangsu Higher Education Institutions
文摘In order to analyze the origin of carbon monoxide(CO) in coal seams, stress–strain experiments under temperature of 50, 150 and 250 °C were conducted using lignite from Kailuan mining area. Fourier transform infrared spectroscopy and elemental analysis were carried out before and after deformation of the samples. The results indicated that CO generated at 150 and 250 °C; the gas component was mostly oxygen(O_2), with small amount of carbon dioxide(CO_2), methane(CH_4) and hydrogen(H_2). At 50 °C, O_2 and a little CO_2 were observed and no CO was found. The carbon content of the coal samples increased slightly after deformation, and the oxygen content, H/C ratio, and O/C ratio decreased. The molecular structure of coal displayed different evolution characteristics at various temperatures. At 50 and 150 °C, the falling off of side chains, broken of ether bond and directional realignment of the aliphatic chains resulting in the formation of long chains were the main performance of coal molecular structure evolution. While at 250 °C, the side chains fell off and short chains formed. Furthermore, at both 150 and 250 °C, condensed degree of aromatic ring increased. Under the action of temperature and pressure, CO forms in two ways.The first is that ether bond breaks, oxygen and carbon atoms combine together and forms CO, or O_2 forming in the broken of ether–oxygen bond leads to the oxidation of free radicals and resulting in the formation of CO. And the second is that CO derives from falling off of C=O group.
基金Project supported by the National Basic Research Program of China(Grant No.2013CB922200)the National Natural Science Foundation of China(Grant Nos.11274141,11034003,11304116,61275128,11247024,and 11274001)the Research Foundation for Basic Research of Jilin Province,China(Grant No.20140101168JC)
文摘Using Bohmlan trajectory （151） method, we investigate the dynamic interference in nlgn-orcter narmonlc generauon from diatomic molecular ions. It is demonstrated that the main characteristics of the molecular harmonic spectrum can be well reproduced by only two BTs which are located at the two ions. This haiapens because these two localized trajectories can receive and store the whole collision information coming from all of the other re-collision trajectories. Therefore, the amplitudes and frequencies of these two trajectories represent the intensity and frequency distribution of the harmonic generation. Moreover, the interference between these two trajectories shows a dip in the harmonic spectrum, which reveals the molecular structure information.
基金Supported by the National Natural Science Foundation of China under Grant No 11404204the Key Project of the Ministry of Education of China under Grant No 211025+1 种基金the Research Fund for the Doctoral Program of Higher Education of China under Grant No 20111404120004the Natural Science Foundation for Young Scientists of Shanxi Province under Grant No 2009021005
文摘We propose an efficient method for the generation of an isolated attosecond pulse from the asymmetric molecular ions HeH^2＋ by adding a half-cycle-like field （HCLF） to the fundamental driving laser field. The high-order harmonic generation （HHG） is investigated by numerically sowing the time-dependent Schrodinger equation. By performing the time-frequency distributions and the electronic wave packet probability densities, we find that the optimizing combined field is not only useful for extending the HHG cutoff, but also for simplifying the recombination channels through controlling the electron localization. In addition, by adjusting the intensity of the HCLF, a dominant short quantum path is selected to contribute the HHG spectrum. As a result, a 75-as isolated attosecond pulse is obtained by superposing a proper range of the harmonics.
基金supported by the Semiconductor Research Corporation (SRC contract No.2012-KJ-2282)
文摘Polyimides are widely used as chip passivation layers and organic substrates in microelectronic packaging. Plasma treatment has been used to enhance the interfacial properties of polyimides, but its molecular mechanism is not clear. In this research, the effects of polyimide surface plasma treatment on the molecular structures at corresponding polyimide/air and buried polyimideJepoxy interfaces were investigated in situ using sum frequency generation （SFG） vibrational spectroscopy. SFG results show that the polyimide backbone molecular structure was different at polyimide/air and polyimideJepoxy interfaces before and after plasma treatment. The different molecular structures at each interface indicate that structural reordering of the polyimide backbone occurred as a result of plasma treatment and contact with the epoxy adhesive. Furthermore, quantitative orientation analysis indicated that plasma treatment of polyimide surfaces altered the twist angle of the polyimide backbone at corresponding buried polyimide/epoxy interfaces. These SFG results indicate that plasma treatment of polymer surfaces can alter the molecular structure at corresponding polymer/air and buried polymer interfaces.
基金supported by the Fundamental Research Funds for the Central Universities(Grant Nos.GK202207012 and QCYRCXM-2022-241).
文摘Permanent dipole moments induced high-order harmonic generation(HHG)signals offer a potential approach to producing elliptically or even circularly polarized X-ray attosecond sources.Previous studies on this topic have mainly focused on diatomic molecules such as CO and HeH.Based on this scheme,significant HHG signals in the direction perpendicular to the molecular axis can be observed in both the high-energy and low-energy regions.However,we found that the high-order harmonics induced by the permanent dipole moments of polyatomic complex molecules involve more intricate physical processes.Using time-dependent density functional theory,we simulated the dynamics of HHG from NH2COOH and NH2COSH interacting with linearly polarized lasers.We found that the harmonic signals in the direction perpendicular to the N-C bond were significantly enhanced in the high-energy photon region.Our analysis indicates that this is due to the complex molecular configuration of NH_(2)COOH and NH_(2)COSH:while the NH_(2) group has C_(2v) symmetry,both COOH and COSH groups lack this symmetry.This structural characteristic results in permanent dipole moments being felt only when electrons return to either COSH or COOH groups,but not to NH_(2) group.Additionally,our results reveal a multi-plateau structure in HHG signal along laser polarization direction,a phenomenon arising from multi-electron and multiorbital effects during interaction between complex molecule and strong laser field.
文摘Current biodegradation timelines show that polyesters take over 200 years to break down. A crucial component of several industries, polyesters are relied upon for materials development and thus require sustainable alternatives. Recent works in generative modeling have made it possible to produce large sets of chemical structures, but current molecular screening methods are expensive, not scalable, and are oversimplified. This work evaluates whether a molecule’s biodegradability potential can be accurately predicted by training a model on recent experimental data. Additionally, three chemical descriptors were evaluated on the final molecules for their effects on biodegradability: molecular structure, bond types, and solubility. A Gradient Boosted Machine was trained on a dataset of 600 molecules and their binary labels on biodegradability. The classification model effectively captured the biodegradability property, yielding an Area Under the Receiver Operating Characteristics, AUROC, of 84% and an Area Under the Precision Recall Curve, or AUPRC, of 87%. Additionally, an existing amortized synthetic tree generation model, SynNet, validated each molecule by showing chemical synthesizability and producing simple and interpretable synthesis pathways. This approach of filtering by prediction and chemical rule interpretation is inexpensive, highly scalable and can capture the necessary complexity. Using this method, novel polyester candidates can be polymerized and produced into sustainable fabrics, reducing environmental stress from textile-reliant industries.
文摘Molecular graph generation is an emerging area of research withnumerous applications. Most existing approaches generate new molecules bylearning a generative model from a collection of known molecules. However,how to obtain the molecular structure from its atomic composition is a problemto be solved. In this paper, it proposes an autoencoder-based model called A2Strwhich takes the atomic composition of a molecular as input. Its molecular structurewas generated and model structures were designed. The effects were verified oneby one. In addition, experiments were implemented with isomer generation anddetermined that the ratio of more than 3% of molecules with different structurescan be obtained from the same combination of atoms.
基金Project supported by the National Natural Science Foundation of China(Grant No.11504151)the Doctoral Scientific Research Foundation of Liaoning Province,China(Grant No.201501123)the Scientific Research Foundation of Liaoning Provincial Education Department,China(Grant No.L2014242)
文摘The laser phase effect on the spatial distribution of the molecular high-order harmonic generation（MHHG） spectrum from H_2~＋ is theoretically investigated through solving the Non-Bohn-Oppenheimer（NBO） time-dependent Schrodinger equation（TDSE）.The results are shown as follows,（i） The generated harmonics from the two nuclei each present an asymmetric distribution.Particularly,when the laser phases are chosen from 0.0π to 0.6π and from 1.7π to 2.0π,the contribution from the negative-H plays a main role in harmonic generation.When the laser phases are chosen from 0.7πto 1.6k,the contribution from the positive-H to the harmonic generation is remarkably enhanced and becomes greater than that from the negative-H.The electron localization,the time-frequency analyses of the harmonic spectrum and the time-dependent wave function are shown to explain the asymmetric harmonic distribution in H_2~＋,which provides us with a method to control the electron motion in molecules,（ⅱ） As the pulse duration increases,the asymmetric distributions of the MHHG in two H nuclei decrease,（ⅲ） Isotope investigation shows that the asymmetric harmonic distribution can be reduced by introducing the heavy nucleus（i.e.,D_2~＋）.
基金supported by the National Natural Science Foundation of China (Nos. 20971045, 21271073)
文摘A new material for both magnetic coupling and electrocatalytic hydrogen generation based on a copper complex,[（HL）CuCl-CuCl（HL）]HCl 1 is prepared by the reaction of 2-（pyridylmethyl）amino-N,N-bis（2-methylene-4,6-difluorophenol）（H2L） and CuCl2·2H2O.In solid,complex 1 is built from two copper units（[（HL）CuCl]）,and exhibits an antiferromagnetic exchange interaction between copper（Ⅱ） ions（J=-160cm^-1）.In liquid,1 can electrocatalyze hydrogen generation both from acetic acid with a turnover frequency（TOF） of 16.3 moles of hydrogen per mole of catalyst per hour at an overpotential（OP）of 941.6 mV（in DMF）,and a neutral buffer with a TOF of 1415.6 moles of hydrogen per mole of catalyst per hour at an OP of 787.6 mV.
基金the National Key Research and Development Program of China under Grant No.2016YFB1000904the National Natural Science Foundation of China under Grant Nos.61922073 and U20A20229the Youth Innovation Promotion Association of Chinese Academy of Sciences under Grant No.2014299。
文摘Generating molecules with desired properties is an important task in chemistry and pharmacy.An efficient method may have a positive impact on finding drugs to treat diseases like COVID-19.Data mining and artificial intelligence may be good ways to find an efficient method.Recently,both the generative models based on deep learning and the work based on genetic algorithms have made some progress in generating molecules and optimizing the molecule’s properties.However,existing methods have defects in the experimental evaluation standards.These methods also need to be improved in efficiency and performance.To solve these problems,we propose a method named the Chemical Genetic Algorithm for Large Molecular Space(CALM).Specifically,CALM employs a scalable and efficient molecular representation called molecular matrix.And we design corresponding crossover,mutation,and mask operators inspired by domain knowledge and previous studies.We apply our genetic algorithm to several tasks related to molecular property optimization and constraint molecular optimization.The results of these tasks show that our approach outperforms the other state-of-the-art deep learning and genetic algorithm methods,where the z tests performed on the results of several experiments show that our method is more than 99%likely to be significant.At the same time,based on the experimental results,we point out the defects in the experimental evaluation standard which affects the fair evaluation of all previous work.Avoiding these defects helps to objectively evaluate the performance of all work.
基金supported by the National Natural Science Fund of China(Grant Nos.12204175,1203400812250003,and 92150105)the Shanghai Sailing Program(Grant No.22YF1411200)+1 种基金the Chenguang Program of Shanghai Education Development Foundation and Shanghai Municipal Education Commission(Grant No.22CGA25)the Science and Technology Commission of Shanghai Municipality(Grant No.21ZR1420100)。
文摘Advances in producing tailored ultrashort laser pulses have enabled the generation and control of molecular dissociative Rydberg excitation along the polarization axis of the laser field.Here,we exploit the orthogonally polarized two-color femtosecond laser fields and achieve an unprecedented two-dimensional control of Rydberg fragment emission in the dissociative frustrated single ionization of oxygen.The Rydberg fragments are collected over the 4πsolid angle,whose momentum distribution is manifested in a characteristic four-lobe pattern.Through precise scanning of the relative phase of the orthogonal two-color laser fields,we demonstrate control over asymmetric directional emission of the Rydberg fragments.Our experimental findings are well supported by classical trajectory Monte Carlo simulations,which suggest an efficient emission control achieved through the manipulation of charge localization upon ionization.
