A significant demand rises for energy-efficient deep neural networks to support power-limited embedding devices with successful deep learning applications in IoT and edge computing fields.An accurate energy prediction...A significant demand rises for energy-efficient deep neural networks to support power-limited embedding devices with successful deep learning applications in IoT and edge computing fields.An accurate energy prediction approach is critical to provide measurement and lead optimization direction.However,the current energy prediction approaches lack accuracy and generalization ability due to the lack of research on the neural network structure and the excessive reliance on customized training dataset.This paper presents a novel energy prediction model,NeurstrucEnergy.NeurstrucEnergy treats neural networks as directed graphs and applies a bi-directional graph neural network training on a randomly generated dataset to extract structural features for energy prediction.NeurstrucEnergy has advantages over linear approaches because the bi-directional graph neural network collects structural features from each layer's parents and children.Experimental results show that NeurstrucEnergy establishes state-of-the-art results with mean absolute percentage error of 2.60%.We also evaluate NeurstrucEnergy in a randomly generated dataset,achieving the mean absolute percentage error of 4.83%over 10 typical convolutional neural networks in recent years and 7 efficient convolutional neural networks created by neural architecture search.Our code is available at https://github.com/NEUSoftGreenAI/NeurstrucEnergy.git.展开更多
In plants,glycerol-3-phosphate dehydrogenase(GPDH)catalyzes the interconversion of glycerol-3-phosphate(G3P)and dihydroxyacetone phosphate(DHAP)coupled to the reduction/oxidation of the nicotinamide adenine dinucleoti...In plants,glycerol-3-phosphate dehydrogenase(GPDH)catalyzes the interconversion of glycerol-3-phosphate(G3P)and dihydroxyacetone phosphate(DHAP)coupled to the reduction/oxidation of the nicotinamide adenine dinucleotide(NADH)pool,and plays a central role in glycerolipid metabolism and stress response.Previous studies have focused mainly on the NAD+-dependent GPDH isoforms,neglecting the role of flavin adenine dinucleotide(FAD)-dependent GPDHs.We isolated and characterized three mitochondrialtargeted FAD-GPDHs in soybean,of which one isoform(GmGPDH12)showed a significant transcriptional response to NaCl and mannitol treatments,suggesting the existence of a major FAD-GPDH isoform acting in soybean responses to salt and osmotic stress.An enzyme kinetic assay showed that the purified GmGPDH12 protein possessed the capacity to oxidize G3P to DHAP in the presence of FAD.Overexpression and RNA interference of GmGPDH12 in soybean hairy roots resulted in elevated tolerance and sensitivity to salt and osmotic stress,respectively.G3P contents were significantly lower in GmGPDH12-overexpressing hair roots and higher in knockdown hair roots,indicating that GmGPDH12 was essential for G3P catabolism.A significant perturbation in redox status of NADH,ascorbic acid(ASA)and glutathione(GSH)pools was observed in GmGPDH12-knockdown plants under stress conditions.The impaired redox balance was manifested by higher reactive oxygen species generation and consequent cell damage or death;however,overexpressing plants showed the opposite results for these traits.GmGPDH12 overexpression contributed to maintaining constant respiration rates under salt or osmotic stress by regulating mRNA levels of key mitochondrial respiratory enzymes.This study provides new evidence for the roles of mitochondria-localized GmGPDH12 in conferring resistance to salt or osmotic stress by maintaining cellular redox homeostasis,protecting cells and respiration from oxidative injury.展开更多
We investigate systematically the effects of the inter-particle attraction on the structure and dynamical behaviors of glass-forming liquids via molecular dynamics simulations.We find that the inter-particle attractio...We investigate systematically the effects of the inter-particle attraction on the structure and dynamical behaviors of glass-forming liquids via molecular dynamics simulations.We find that the inter-particle attraction does not influence the structure,but greatly affects the dynamics and dynamical heterogeneity of the system.After the system changes from a purely repulsive glass-forming liquid to an attractive one,the dynamics slows down and the dynamical heterogeneity becomes greater,which is found interestingly to be associated with larger cooperative rearrangement regions(CRRs).Additionally,the structures of CRRs are observed to be compact in attractive glass-forming liquids but string-like in purely repulsive ones.Our findings constitute an important contribution to the ongoing study of the role of attractions in properties of glasses and glass-forming liquids.展开更多
Identification of the glass formation process in various conditions is of importance for fundamental understanding of the mechanism of glass transitions as well as for developments and applications of glassy materials...Identification of the glass formation process in various conditions is of importance for fundamental understanding of the mechanism of glass transitions as well as for developments and applications of glassy materials.We investigate the role of pinning in driving the transformation of crystal into glass in two-dimensional colloidal suspensions of monodisperse microspheres.The pinning is produced by immobilizing a fraction of microspheres on the substrate of sample cells where the mobile microspheres sediment.Structurally,the crystal-hexatic-glass transition occurs with increasing the number fraction of pinningρpinning,and the orientational correlation exhibits a change from quasi-long-range to short-range order atρpinning=0.02.Interestingly,the dynamics shows a nonmonotonic change with increasing the fraction of pinning.This is due to the competition between the disorder that enhances the dynamics and the pinning that hinders the particle motions.Our work highlights the important role of the pinning on the colloidal glass transition,which not only provides a new strategy to prevent crystallization forming glass,but also is helpful for understanding of the vitrification in colloidal systems.展开更多
Noise-like pulses having a pedestal of 690 fs and a spike of 59.6 fs were generated in a nonlinear Yb-doped fiber amplification system. The seed source is a mode-locked Yb-doped fiber laser by nonlinear polarization r...Noise-like pulses having a pedestal of 690 fs and a spike of 59.6 fs were generated in a nonlinear Yb-doped fiber amplification system. The seed source is a mode-locked Yb-doped fiber laser by nonlinear polarization rotation,and dissipative soliton pulses were obtained in it. Then, the dissipative soliton pulses passed through a 7.6 m dispersive fiber to enhance the dispersion and nonlinearity. Further on, the dissipative soliton pulses were launched into a Yb-doped fiber nonlinear amplifier, and stable noise-like pulses with a pedestal of 6.26 ps and a spike of 227 fs were achieved. Finally, by a grating pair, the pedestal and spike of the noise-like pulses were effectively compressed to 690 fs and 59.6 fs, respectively. To the best of our knowledge, this is the shortest pedestal demonstrated in noise-like pulses operating at 1 μm.展开更多
The dispersive Fourier transform technique provides feasibility of exploring non-repetitive events and the buildup process in ultrafast lasers.In this paper,we report a new buildup process of dissipative solitons in a...The dispersive Fourier transform technique provides feasibility of exploring non-repetitive events and the buildup process in ultrafast lasers.In this paper,we report a new buildup process of dissipative solitons in a simplified mode-locked Yb-doped fiber laser,which includes more complex physics stages such as the Q-switching stage,raised and damped relaxation oscillation stages,noise-like stage,successive soliton explosions stage,and soliton breathing stage.Complete evolution dynamics of noise-like pulse and double pulse are also investigated with dispersive Fourier transform.For the noise-like pulse dynamics process,it will only experience the Q-switching and relaxation oscillation stages.In the case of dissipative soliton and noise-like pulse,the double pulse buildup behavior is manifested as the replication of individual pulses.A weak energy migration occurs between two pulses before reaching steady state.Meanwhile,real-time mutual conversion of the dissipative soliton and noise-like pulse has been experimentally observed,which appears to be instantaneous without extra physical processes.To the best of our knowledge,this is the first report on these physical phenomena observed together in a mode-locked fiber laser.The results further enrich the dynamics of mode-locked fiber lasers and provide potential conditions for obtaining intelligent mode-locked lasers with controllable output.展开更多
Targeting of chemotherapeutics towards a tumor site by magnetic nanocarriers is considered promising in tumor-control.Magnetic nanoparticles are also considered for use in infection-control as a new means to prevent a...Targeting of chemotherapeutics towards a tumor site by magnetic nanocarriers is considered promising in tumor-control.Magnetic nanoparticles are also considered for use in infection-control as a new means to prevent antimicrobial resistance from becoming the number one cause of death by the year 2050.To this end,magnetic nanoparticles can either be loaded with an antimicrobial for use as a delivery vehicle or modified to acquire intrinsic antimicrobial properties.Magnetic nanoparticles can also be used for the local generation of heat to kill infectious microorganisms.Although appealing for tumor-and infectioncontrol,injection in the blood circulation may yield reticuloendothelial uptake and physical obstruction in organs that yield reduced targeting efficiency.This can be prevented with suitable surface modification.However,precise techniques to direct magnetic nanoparticles towards a target site are lacking.The problem of precise targeting is aggravated in infection-control due to the micrometer-size of infectious biofilms,as opposed to targeting of nanoparticles towards centimeter-sized tumors.This review aims to identify possibilities and impossibilities of magnetic targeting of nanoparticles for infection-control.We first review targeting techniques and the spatial resolution they can achieve as well as surface-chemical modifications of magnetic nanoparticles to enhance their targeting efficiency and antimicrobial efficacy.It is concluded that targeting problems encountered in tumor-control using magnetic nanoparticles,are neglected in most studies on their potential application in infection-control.Currently biofilm targeting by smart,self-adaptive and pH-responsive,antimicrobial nanocarriers for instance,seems easier to achieve than magnetic targeting.This leads to the conclusion that magnetic targeting of nanoparticles for the control of micrometer-sized infectious biofilms may be less promising than initially expected.However,using propulsion rather than precise targeting of magnetic nanoparticles in a magnetic field to traverse through infectious-biofilms can create artificial channels for enhanced antibiotic transport.This is identified as a more feasible,innovative application of magnetic nanoparticles in infection-control than precise targeting and distribution of magnetic nanoparticles over the depth of a biofilm.展开更多
基金supported by the Natural Science Foundation of Liaoning Province(2020-BS-054)the Fundamental Research Funds for the Central Universities(N2017005)the National Natural Science Foundation of China(62162050).
文摘A significant demand rises for energy-efficient deep neural networks to support power-limited embedding devices with successful deep learning applications in IoT and edge computing fields.An accurate energy prediction approach is critical to provide measurement and lead optimization direction.However,the current energy prediction approaches lack accuracy and generalization ability due to the lack of research on the neural network structure and the excessive reliance on customized training dataset.This paper presents a novel energy prediction model,NeurstrucEnergy.NeurstrucEnergy treats neural networks as directed graphs and applies a bi-directional graph neural network training on a randomly generated dataset to extract structural features for energy prediction.NeurstrucEnergy has advantages over linear approaches because the bi-directional graph neural network collects structural features from each layer's parents and children.Experimental results show that NeurstrucEnergy establishes state-of-the-art results with mean absolute percentage error of 2.60%.We also evaluate NeurstrucEnergy in a randomly generated dataset,achieving the mean absolute percentage error of 4.83%over 10 typical convolutional neural networks in recent years and 7 efficient convolutional neural networks created by neural architecture search.Our code is available at https://github.com/NEUSoftGreenAI/NeurstrucEnergy.git.
基金financially supported by National Natural Science Foundation of China(31701449,31971968,31971899,31501332)National Key Research and Development Program of China(2016YFD 0100500,2016YFD0100300,2016YFD0100201-21,JFYS2016ZY03003792-01-21)+7 种基金China Postdoctoral Science Foundation(2019M661243)Postdoctoral Project of Northeast Agricultural University(NEAUBH-19002)EUCLEG(727312,2017YFE0111000)Natural Science Foundation of Heilongjiang Province(QC2017013)Special Financial Aid to Post-doctor Research Fellow in Heilongjiang(LBH-TZ1714)Heilongjiang Academy of Agricultural Sciences Funds(2019YYYF019)International Postdoctoral Exchange Fellowship Program of China Postdoctoral Council(20180004)Heilongjiang Funds for Distinguished Young Scientists(JC2016004,JC2017006)。
文摘In plants,glycerol-3-phosphate dehydrogenase(GPDH)catalyzes the interconversion of glycerol-3-phosphate(G3P)and dihydroxyacetone phosphate(DHAP)coupled to the reduction/oxidation of the nicotinamide adenine dinucleotide(NADH)pool,and plays a central role in glycerolipid metabolism and stress response.Previous studies have focused mainly on the NAD+-dependent GPDH isoforms,neglecting the role of flavin adenine dinucleotide(FAD)-dependent GPDHs.We isolated and characterized three mitochondrialtargeted FAD-GPDHs in soybean,of which one isoform(GmGPDH12)showed a significant transcriptional response to NaCl and mannitol treatments,suggesting the existence of a major FAD-GPDH isoform acting in soybean responses to salt and osmotic stress.An enzyme kinetic assay showed that the purified GmGPDH12 protein possessed the capacity to oxidize G3P to DHAP in the presence of FAD.Overexpression and RNA interference of GmGPDH12 in soybean hairy roots resulted in elevated tolerance and sensitivity to salt and osmotic stress,respectively.G3P contents were significantly lower in GmGPDH12-overexpressing hair roots and higher in knockdown hair roots,indicating that GmGPDH12 was essential for G3P catabolism.A significant perturbation in redox status of NADH,ascorbic acid(ASA)and glutathione(GSH)pools was observed in GmGPDH12-knockdown plants under stress conditions.The impaired redox balance was manifested by higher reactive oxygen species generation and consequent cell damage or death;however,overexpressing plants showed the opposite results for these traits.GmGPDH12 overexpression contributed to maintaining constant respiration rates under salt or osmotic stress by regulating mRNA levels of key mitochondrial respiratory enzymes.This study provides new evidence for the roles of mitochondria-localized GmGPDH12 in conferring resistance to salt or osmotic stress by maintaining cellular redox homeostasis,protecting cells and respiration from oxidative injury.
基金Project supported by the National Natural Science Foundation of China (Grant Nos. 11704270,12074275,and 11574222)Jiangsu Planned Projects for Postdoctoral Research Funds+1 种基金the PAPD Program of Jiangsu Higher Education Institutionsthe Start-up Fund from Anhui University (Grant No. S020318001/02)
文摘We investigate systematically the effects of the inter-particle attraction on the structure and dynamical behaviors of glass-forming liquids via molecular dynamics simulations.We find that the inter-particle attraction does not influence the structure,but greatly affects the dynamics and dynamical heterogeneity of the system.After the system changes from a purely repulsive glass-forming liquid to an attractive one,the dynamics slows down and the dynamical heterogeneity becomes greater,which is found interestingly to be associated with larger cooperative rearrangement regions(CRRs).Additionally,the structures of CRRs are observed to be compact in attractive glass-forming liquids but string-like in purely repulsive ones.Our findings constitute an important contribution to the ongoing study of the role of attractions in properties of glasses and glass-forming liquids.
基金Supported by the National Natural Science Foundation of China (Grant Nos. 12074275, 11704269, and 11704270)the Natural Science Foundation of the Jiangsu Higher Education Institutions of China (Grant Nos. 20KJA150008 and 17KJB140020)the PAPD Program of Jiangsu Higher Education Institutions
文摘Identification of the glass formation process in various conditions is of importance for fundamental understanding of the mechanism of glass transitions as well as for developments and applications of glassy materials.We investigate the role of pinning in driving the transformation of crystal into glass in two-dimensional colloidal suspensions of monodisperse microspheres.The pinning is produced by immobilizing a fraction of microspheres on the substrate of sample cells where the mobile microspheres sediment.Structurally,the crystal-hexatic-glass transition occurs with increasing the number fraction of pinningρpinning,and the orientational correlation exhibits a change from quasi-long-range to short-range order atρpinning=0.02.Interestingly,the dynamics shows a nonmonotonic change with increasing the fraction of pinning.This is due to the competition between the disorder that enhances the dynamics and the pinning that hinders the particle motions.Our work highlights the important role of the pinning on the colloidal glass transition,which not only provides a new strategy to prevent crystallization forming glass,but also is helpful for understanding of the vitrification in colloidal systems.
基金the Beijing Natural Science Foundation(No.4192015)the National Natural Science Foundation of China(No.61975003).
文摘Noise-like pulses having a pedestal of 690 fs and a spike of 59.6 fs were generated in a nonlinear Yb-doped fiber amplification system. The seed source is a mode-locked Yb-doped fiber laser by nonlinear polarization rotation,and dissipative soliton pulses were obtained in it. Then, the dissipative soliton pulses passed through a 7.6 m dispersive fiber to enhance the dispersion and nonlinearity. Further on, the dissipative soliton pulses were launched into a Yb-doped fiber nonlinear amplifier, and stable noise-like pulses with a pedestal of 6.26 ps and a spike of 227 fs were achieved. Finally, by a grating pair, the pedestal and spike of the noise-like pulses were effectively compressed to 690 fs and 59.6 fs, respectively. To the best of our knowledge, this is the shortest pedestal demonstrated in noise-like pulses operating at 1 μm.
基金This work was supported by the Beijing Natural Science Foundation(No.4192015)the National Natural Science Foundation of China(No.61975003).
文摘The dispersive Fourier transform technique provides feasibility of exploring non-repetitive events and the buildup process in ultrafast lasers.In this paper,we report a new buildup process of dissipative solitons in a simplified mode-locked Yb-doped fiber laser,which includes more complex physics stages such as the Q-switching stage,raised and damped relaxation oscillation stages,noise-like stage,successive soliton explosions stage,and soliton breathing stage.Complete evolution dynamics of noise-like pulse and double pulse are also investigated with dispersive Fourier transform.For the noise-like pulse dynamics process,it will only experience the Q-switching and relaxation oscillation stages.In the case of dissipative soliton and noise-like pulse,the double pulse buildup behavior is manifested as the replication of individual pulses.A weak energy migration occurs between two pulses before reaching steady state.Meanwhile,real-time mutual conversion of the dissipative soliton and noise-like pulse has been experimentally observed,which appears to be instantaneous without extra physical processes.To the best of our knowledge,this is the first report on these physical phenomena observed together in a mode-locked fiber laser.The results further enrich the dynamics of mode-locked fiber lasers and provide potential conditions for obtaining intelligent mode-locked lasers with controllable output.
基金the National Key Research and Development Program of China(No.2016YFC1100402)the National Natural Science Foundation of China(Nos.11574222 and 21522404)the University Medical Center Groningen(UMCG),The Netherlands。
文摘Targeting of chemotherapeutics towards a tumor site by magnetic nanocarriers is considered promising in tumor-control.Magnetic nanoparticles are also considered for use in infection-control as a new means to prevent antimicrobial resistance from becoming the number one cause of death by the year 2050.To this end,magnetic nanoparticles can either be loaded with an antimicrobial for use as a delivery vehicle or modified to acquire intrinsic antimicrobial properties.Magnetic nanoparticles can also be used for the local generation of heat to kill infectious microorganisms.Although appealing for tumor-and infectioncontrol,injection in the blood circulation may yield reticuloendothelial uptake and physical obstruction in organs that yield reduced targeting efficiency.This can be prevented with suitable surface modification.However,precise techniques to direct magnetic nanoparticles towards a target site are lacking.The problem of precise targeting is aggravated in infection-control due to the micrometer-size of infectious biofilms,as opposed to targeting of nanoparticles towards centimeter-sized tumors.This review aims to identify possibilities and impossibilities of magnetic targeting of nanoparticles for infection-control.We first review targeting techniques and the spatial resolution they can achieve as well as surface-chemical modifications of magnetic nanoparticles to enhance their targeting efficiency and antimicrobial efficacy.It is concluded that targeting problems encountered in tumor-control using magnetic nanoparticles,are neglected in most studies on their potential application in infection-control.Currently biofilm targeting by smart,self-adaptive and pH-responsive,antimicrobial nanocarriers for instance,seems easier to achieve than magnetic targeting.This leads to the conclusion that magnetic targeting of nanoparticles for the control of micrometer-sized infectious biofilms may be less promising than initially expected.However,using propulsion rather than precise targeting of magnetic nanoparticles in a magnetic field to traverse through infectious-biofilms can create artificial channels for enhanced antibiotic transport.This is identified as a more feasible,innovative application of magnetic nanoparticles in infection-control than precise targeting and distribution of magnetic nanoparticles over the depth of a biofilm.