In this paper,we propose mesoscience-guided deep learning(MGDL),a deep learning modeling approach guided by mesoscience,to study complex systems.When establishing sample dataset based on the same system evolution data...In this paper,we propose mesoscience-guided deep learning(MGDL),a deep learning modeling approach guided by mesoscience,to study complex systems.When establishing sample dataset based on the same system evolution data,different from the operation of conventional deep learning method,MGDL introduces the treatment of the dominant mechanisms of complex system and interactions between them according to the principle of compromise in competition(CIC)in mesoscience.Mesoscience constraints are then integrated into the loss function to guide the deep learning training.Two methods are proposed for the addition of mesoscience constraints.The physical interpretability of the model-training process is improved by MGDL because guidance and constraints based on physical principles are provided.MGDL was evaluated using a bubbling bed modeling case and compared with traditional techniques.With a much smaller training dataset,the results indicate that mesoscience-constraint-based model training has distinct advantages in terms of convergence stability and prediction accuracy,and it can be widely applied to various neural network configurations.The MGDL approach proposed in this paper is a novel method for utilizing the physical background information during deep learning model training.Further exploration of MGDL will be continued in the future.展开更多
The paper“Challenges in Immune System:Mesoscale and mesoregime complexity”by Ren et al.[1]represents more than an important advance in immunology;it offers a growing recognition of mesoscience as a unifying scientif...The paper“Challenges in Immune System:Mesoscale and mesoregime complexity”by Ren et al.[1]represents more than an important advance in immunology;it offers a growing recognition of mesoscience as a unifying scientific framework for understanding complexity.展开更多
Electrocatalytic materials with different morphologies,sizes,and components show different catalytic behavior in various heterogeneous catalytic reactions.It has been proved that the catalytic properties of these mate...Electrocatalytic materials with different morphologies,sizes,and components show different catalytic behavior in various heterogeneous catalytic reactions.It has been proved that the catalytic properties of these materials are strongly influenced by several factors at different levels,including the electrode morphology,reaction channels,three-phase interface,and surface active sites.Recent developments of mesoscience allow one to study the relationship between the apparent catalytic performance of electro-catalytic materials with these factors from different levels.In this review,following a brief introduction of new mesoscience,we summarize the effect of mesoscience on electrocatalytic material design,including modulating the geometric and electronic structures of materials focusing on morphology(particulate,fiber,film,array,monolith,and superlattice),pore structure(microporous,mesoporous,and hierarchical),size(single atoms,nanoclusters,and nanoparticles),multiple components(alloys,heterostructures,and multiple ligands),and crystal structures(crystalline,amorphous,and multiple crystal phases).By evaluating the electrocatalytic performance of catalytic materials tuned at the mesoscale,we paint a picture of how these factors at different levels affect the final system performance and then provide a new direction to better understand and design catalytic materials from the viewpoint of mesoscience.展开更多
Mesoscale characteristics and their interdimensional correlation are the focus of contemporary interdisciplinary research.Mesoscience is a discipline that has the potential to radically update the existing knowledge s...Mesoscale characteristics and their interdimensional correlation are the focus of contemporary interdisciplinary research.Mesoscience is a discipline that has the potential to radically update the existing knowledge structure,which differs from the conventional unit-scale and system-scale research models,revealing a previously untouchable area for scientific research.Integrative biology research aims to dissect the complex problems of life systems by conducting comprehensive research and integrating various disciplines from all biological levels of the living organism.However,the mesoscientific issues between different research units are neglected and challenging.Mesoscale research in biology requires the integration of research theories and methods from other disciplines(mathematics,physics,engineering,and even visual imaging)to investigate theoretical and frontier questions of biological processes through experiments,computations,and modeling.We reviewed integrative paradigms and methods for the biological mesoscale problems(focusing on oncology research)and prospected the potential of their multiple dimensions and upcoming challenges.We expect to establish an interactive and collaborative theoretical platform for further expanding the depth and width of our understanding on the nature of biology.展开更多
Exploring the physical mechanisms of complex systems and making effective use of them are the keys to dealing with the complexity of the world.The emergence of big data and the enhancement of computing power,in conjun...Exploring the physical mechanisms of complex systems and making effective use of them are the keys to dealing with the complexity of the world.The emergence of big data and the enhancement of computing power,in conjunction with the improvement of optimization algorithms,are leading to the development of artificial intelligence(AI)driven by deep learning.However,deep learning fails to reveal the underlying logic and physical connotations of the problems being solved.Mesoscience provides a concept to understand the mechanism of the spatiotemporal multiscale structure of complex systems,and its capability for analyzing complex problems has been validated in different fields.This paper proposes a research paradigm for AI,which introduces the analytical principles of mesoscience into the design of deep learning models.This is done to address the fundamental problem of deep learning models detaching the physical prototype from the problem being solved;the purpose is to promote the sustainable development of AI.展开更多
One of the greatest challenges in bioscience is to gain a unified view of the human immune system.This paper presents a perspective on the multilevel and multiscale complexities inherent in the immune system and discu...One of the greatest challenges in bioscience is to gain a unified view of the human immune system.This paper presents a perspective on the multilevel and multiscale complexities inherent in the immune system and discusses how these features influence the traditional research methodologies rooted in reductionism and holism.It is acknowledged that diverse complexities of the immune system are multilevel in nature,with each level showing multiscale properties,and the complexity emerging always at mesoscales in mesoregimes.Mesoscience encompasses both mesoscales,which represent the intermediate scales between the element scales and system scales at different levels,and mesoregimes,defined as transitional regimes governed by the interplay of at least two dominant mechanisms between two limiting regimes at each level.Therefore,mesoscience provides a promising paradigm to study the complexity and diversity of the immune system by bridging the gaps across multiple hierarchical levels.In particular,we focus on the mesoscience methodology to address the complexities in the immune system and offer insights into potential diagnostic,therapeutic,and theranostic strategies for immune‐related diseases from a mesoscience perspective.展开更多
Based on the existing energy-minimization multi-scale(EMMS)model for turbulent flow in pipe,an improved version is proposed,in which not only a new radial velocity distribution is introduced but also the quantificatio...Based on the existing energy-minimization multi-scale(EMMS)model for turbulent flow in pipe,an improved version is proposed,in which not only a new radial velocity distribution is introduced but also the quantification of total dissipation over the cross-section of pipe is improved for the dominant mechanism of fully turbulent flow in pipe.Then four dynamic equality constraints and some other constraints are constructed but there are five parameters involved,leading to one free variable left.Through the compromise in competition between dominant mechanisms for laminar and fully turbulent flow in pipe respectively,the above four constructed dynamic equality constraints can be closed.Finally,the cases for turbulent flow in pipe with low,moderate and high Reynolds number are simulated by the improved EMMS model.The numerical results show that the model can obtain reasonable results which agree well with the data computed by the direct numerical simulation and those obtained by experi-ment.This illustrates that the improved EMMS model for turbulent flow in pipe is reasonable and the compromise in competition between dominant mechanisms is indeed a universal governing principle hidden in complex systems.Especially,one more EMMS model for a complex system is offered,pro-moting the further development of mesoscience.展开更多
基金supported by the National Natural Science Foundation of China(62050226 and 22078327)the International Partnership Program of Chinese Academy of Sciences(122111KYSB20170068).
文摘In this paper,we propose mesoscience-guided deep learning(MGDL),a deep learning modeling approach guided by mesoscience,to study complex systems.When establishing sample dataset based on the same system evolution data,different from the operation of conventional deep learning method,MGDL introduces the treatment of the dominant mechanisms of complex system and interactions between them according to the principle of compromise in competition(CIC)in mesoscience.Mesoscience constraints are then integrated into the loss function to guide the deep learning training.Two methods are proposed for the addition of mesoscience constraints.The physical interpretability of the model-training process is improved by MGDL because guidance and constraints based on physical principles are provided.MGDL was evaluated using a bubbling bed modeling case and compared with traditional techniques.With a much smaller training dataset,the results indicate that mesoscience-constraint-based model training has distinct advantages in terms of convergence stability and prediction accuracy,and it can be widely applied to various neural network configurations.The MGDL approach proposed in this paper is a novel method for utilizing the physical background information during deep learning model training.Further exploration of MGDL will be continued in the future.
文摘The paper“Challenges in Immune System:Mesoscale and mesoregime complexity”by Ren et al.[1]represents more than an important advance in immunology;it offers a growing recognition of mesoscience as a unifying scientific framework for understanding complexity.
基金This research work was financially sponsored by the National Key Research and Development Program of China(2016YFB0101202)the Key Program of the National Nature Science Foundation of China(Grant No.91534205,No.21376283 and No.21576032).
文摘Electrocatalytic materials with different morphologies,sizes,and components show different catalytic behavior in various heterogeneous catalytic reactions.It has been proved that the catalytic properties of these materials are strongly influenced by several factors at different levels,including the electrode morphology,reaction channels,three-phase interface,and surface active sites.Recent developments of mesoscience allow one to study the relationship between the apparent catalytic performance of electro-catalytic materials with these factors from different levels.In this review,following a brief introduction of new mesoscience,we summarize the effect of mesoscience on electrocatalytic material design,including modulating the geometric and electronic structures of materials focusing on morphology(particulate,fiber,film,array,monolith,and superlattice),pore structure(microporous,mesoporous,and hierarchical),size(single atoms,nanoclusters,and nanoparticles),multiple components(alloys,heterostructures,and multiple ligands),and crystal structures(crystalline,amorphous,and multiple crystal phases).By evaluating the electrocatalytic performance of catalytic materials tuned at the mesoscale,we paint a picture of how these factors at different levels affect the final system performance and then provide a new direction to better understand and design catalytic materials from the viewpoint of mesoscience.
基金National Key Research and Development Program of China,Grant/Award Numbers:2022YFE0103600,2021YFF1201300CAMS Innovation Fund for Medical Sciences(CIFMS),Grant/Award Number:2021-I2M-1-014+2 种基金National Natural Science Foundation of China,Grant/Award Numbers:81872280,82073094Open Issue of State Key Laboratory of Molecular Oncology,Grant/Award Number:SKL-KF-2021-16Independent Issue of State Key Laboratory of Molecular Oncology,Grant/Award Number:SKL-2021-16。
文摘Mesoscale characteristics and their interdimensional correlation are the focus of contemporary interdisciplinary research.Mesoscience is a discipline that has the potential to radically update the existing knowledge structure,which differs from the conventional unit-scale and system-scale research models,revealing a previously untouchable area for scientific research.Integrative biology research aims to dissect the complex problems of life systems by conducting comprehensive research and integrating various disciplines from all biological levels of the living organism.However,the mesoscientific issues between different research units are neglected and challenging.Mesoscale research in biology requires the integration of research theories and methods from other disciplines(mathematics,physics,engineering,and even visual imaging)to investigate theoretical and frontier questions of biological processes through experiments,computations,and modeling.We reviewed integrative paradigms and methods for the biological mesoscale problems(focusing on oncology research)and prospected the potential of their multiple dimensions and upcoming challenges.We expect to establish an interactive and collaborative theoretical platform for further expanding the depth and width of our understanding on the nature of biology.
基金We would like to thank Dr.Wenlai Huang,Dr.Jianhua Chen,and Dr.Lin Zhang for the valuable discussionWe thank the editors and reviewers for their valuable comments about this articleWe gratefully acknowledge the support from the National Natural Science Foundation of China(91834303).
文摘Exploring the physical mechanisms of complex systems and making effective use of them are the keys to dealing with the complexity of the world.The emergence of big data and the enhancement of computing power,in conjunction with the improvement of optimization algorithms,are leading to the development of artificial intelligence(AI)driven by deep learning.However,deep learning fails to reveal the underlying logic and physical connotations of the problems being solved.Mesoscience provides a concept to understand the mechanism of the spatiotemporal multiscale structure of complex systems,and its capability for analyzing complex problems has been validated in different fields.This paper proposes a research paradigm for AI,which introduces the analytical principles of mesoscience into the design of deep learning models.This is done to address the fundamental problem of deep learning models detaching the physical prototype from the problem being solved;the purpose is to promote the sustainable development of AI.
基金funded by the National Natural Science Foundation of China(Grant Nos.22373105,T2394501,and T2342011)the Strategic Priority Research Program of the Chinese Academy of Sciences(Grant No.XDA0390501).
文摘One of the greatest challenges in bioscience is to gain a unified view of the human immune system.This paper presents a perspective on the multilevel and multiscale complexities inherent in the immune system and discusses how these features influence the traditional research methodologies rooted in reductionism and holism.It is acknowledged that diverse complexities of the immune system are multilevel in nature,with each level showing multiscale properties,and the complexity emerging always at mesoscales in mesoregimes.Mesoscience encompasses both mesoscales,which represent the intermediate scales between the element scales and system scales at different levels,and mesoregimes,defined as transitional regimes governed by the interplay of at least two dominant mechanisms between two limiting regimes at each level.Therefore,mesoscience provides a promising paradigm to study the complexity and diversity of the immune system by bridging the gaps across multiple hierarchical levels.In particular,we focus on the mesoscience methodology to address the complexities in the immune system and offer insights into potential diagnostic,therapeutic,and theranostic strategies for immune‐related diseases from a mesoscience perspective.
基金supported by the State Key Laboratory of Multiphase Complex Systems(grant No.MPCS-2022-A-01)the National Natural Science Foundation of China(grant No.22078327).
文摘Based on the existing energy-minimization multi-scale(EMMS)model for turbulent flow in pipe,an improved version is proposed,in which not only a new radial velocity distribution is introduced but also the quantification of total dissipation over the cross-section of pipe is improved for the dominant mechanism of fully turbulent flow in pipe.Then four dynamic equality constraints and some other constraints are constructed but there are five parameters involved,leading to one free variable left.Through the compromise in competition between dominant mechanisms for laminar and fully turbulent flow in pipe respectively,the above four constructed dynamic equality constraints can be closed.Finally,the cases for turbulent flow in pipe with low,moderate and high Reynolds number are simulated by the improved EMMS model.The numerical results show that the model can obtain reasonable results which agree well with the data computed by the direct numerical simulation and those obtained by experi-ment.This illustrates that the improved EMMS model for turbulent flow in pipe is reasonable and the compromise in competition between dominant mechanisms is indeed a universal governing principle hidden in complex systems.Especially,one more EMMS model for a complex system is offered,pro-moting the further development of mesoscience.
