Density-functional theory (DFT) is a widespread method for sim- ulating the quantum-chemical behaviour of electrons in matter. It provides a first-principles description of many optical, me- chanical and chemical properties at an acceptable computational cost [16, 2, 3]. For a wide range of systems the obtained predic- tions are accurate and shortcomings of the theory are by now well-understood [2, 3]. The desire to tackle even bigger systems and more involved materials, however, keeps posing novel challenges that require methods to constantly improve. One example are so- called high-throughput screening approaches, which are becoming prominent in recent years. In these techniques one wishes to sys- tematically scan over huge design spaces of compounds in order to identify promising novel materials for targeted follow-up investi- gation. This has already lead to many success stories [14], such as the discovery of novel earth-abundant semiconductors [11], novel light-absorbing materials [20], electrocatalysts [8], materials for hydrogen storage [13] or for Li-ion batteries [1]. Keeping in mind the large range of physics that needs to be covered in these studies as well as the typical number of calculations (up to the order of millions), a bottleneck in these studies is the reliability and performance of the underlying DFT codes. To tackle these aspects multidisciplinary collaboration with mathematicians developing more numerically stable algorithms, computer scientists providing high-performance implementations, physicists and chemists designing appropriate models, and appli-cation scientists integrating the resulting methods inside a suitable simulation workflow is essential. While to date already a size-able number of DFT codes exist, e.g. ABINIT [19], Quantum- Espresso [6] or VASP [15] to name only a few, they lack sufficient flexibility inside their low-level computational routines to easily support fundamental research in computer science or mathematics. To test
{"title":"DFTK: A Julian approach for simulating electrons in solids","authors":"Michael F. Herbst, A. Levitt, É. Cancès","doi":"10.21105/JCON.00069","DOIUrl":"https://doi.org/10.21105/JCON.00069","url":null,"abstract":"Density-functional theory (DFT) is a widespread method for sim- ulating the quantum-chemical behaviour of electrons in matter. It provides a first-principles description of many optical, me- chanical and chemical properties at an acceptable computational cost [16, 2, 3]. For a wide range of systems the obtained predic- tions are accurate and shortcomings of the theory are by now well-understood [2, 3]. The desire to tackle even bigger systems and more involved materials, however, keeps posing novel challenges that require methods to constantly improve. One example are so- called high-throughput screening approaches, which are becoming prominent in recent years. In these techniques one wishes to sys- tematically scan over huge design spaces of compounds in order to identify promising novel materials for targeted follow-up investi- gation. This has already lead to many success stories [14], such as the discovery of novel earth-abundant semiconductors [11], novel light-absorbing materials [20], electrocatalysts [8], materials for hydrogen storage [13] or for Li-ion batteries [1]. Keeping in mind the large range of physics that needs to be covered in these studies as well as the typical number of calculations (up to the order of millions), a bottleneck in these studies is the reliability and performance of the underlying DFT codes. To tackle these aspects multidisciplinary collaboration with mathematicians developing more numerically stable algorithms, computer scientists providing high-performance implementations, physicists and chemists designing appropriate models, and appli-cation scientists integrating the resulting methods inside a suitable simulation workflow is essential. While to date already a size-able number of DFT codes exist, e.g. ABINIT [19], Quantum- Espresso [6] or VASP [15] to name only a few, they lack sufficient flexibility inside their low-level computational routines to easily support fundamental research in computer science or mathematics. To test","PeriodicalId":443465,"journal":{"name":"JuliaCon Proceedings","volume":"109 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117125791","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
D. Panigo, P. Gluzmann, E. Mocskos, Adan Mauri Ungaro, Valentin Mari, Nicolás Monzón
The aim of this paper is twofold. The first one is to describe a novel research-project designed for building bridges between machine learning and econometric worlds ( ModelSelection.jl). The second one is to introduce the main characteristics and comparative performance of the first Julia-native all-subset regression algorithm included in GlobalSearchRegression.jl (v1.0.5). As other available alternatives, this algorithm allows researchers to obtain the best model specification among all possible covariate combinations - in terms of user defined information criteria-, but up to 3165 and 197 times faster than STATA and R alternatives, respectively.
{"title":"GlobalSearchRegression.jl: Building bridges between Machine Learning and Econometrics in Fat-Data scenarios","authors":"D. Panigo, P. Gluzmann, E. Mocskos, Adan Mauri Ungaro, Valentin Mari, Nicolás Monzón","doi":"10.21105/jcon.00053","DOIUrl":"https://doi.org/10.21105/jcon.00053","url":null,"abstract":"The aim of this paper is twofold. The first one is to describe a novel research-project designed for building bridges between machine learning and econometric worlds ( ModelSelection.jl). The second one is to introduce the main characteristics and comparative performance of the first Julia-native all-subset regression algorithm included in GlobalSearchRegression.jl (v1.0.5). As other available alternatives, this algorithm allows researchers to obtain the best model specification among all possible covariate combinations - in terms of user defined information criteria-, but up to 3165 and 197 times faster than STATA and R alternatives, respectively.","PeriodicalId":443465,"journal":{"name":"JuliaCon Proceedings","volume":"88 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125105904","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Econometrics.jl is a package for econometrics analysis. It provides a series of most common routines for applied econometrics such as models for continuous, nominal, and ordinal outcomes, longitudinal estimators, variable absorption, and support for convenience functionality such as weights, rank deficient, and robust variance covariance estimators. This study complements the package through a discussion of the motivation, placing the contribution within the Julia ecosystem and econometrics software in general, and provides insights on current gaps and ways the Julia ecosystem can evolve.
{"title":"Econometrics.jl","authors":"J. Calderón","doi":"10.21105/jcon.00038","DOIUrl":"https://doi.org/10.21105/jcon.00038","url":null,"abstract":"Econometrics.jl is a package for econometrics analysis. It provides a series of most common routines for applied econometrics such as models for continuous, nominal, and ordinal outcomes, longitudinal estimators, variable absorption, and support for convenience functionality such as weights, rank deficient, and robust variance covariance estimators. This study complements the package through a discussion of the motivation, placing the contribution within the Julia ecosystem and econometrics software in general, and provides insights on current gaps and ways the Julia ecosystem can evolve.","PeriodicalId":443465,"journal":{"name":"JuliaCon Proceedings","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129102925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ranjan Anantharaman, K. Hall, Viral B. Shah, A. Edelman
Connectivity across landscapes influences a wide range of conservation-relevant ecological processes, including species movements, gene flow, and the spread of wildfire, pests, and diseases. Recent improvements in remote sensing data suggest great potential to advance connectivity models, but computational constraints hinder these advances. To address this challenge, we upgraded the widely-used Circuitscape connectivity package to the high performance Julia programming language. Circuitscape.jl allows users to solve problems faster via improved parallel processing and solvers, and supports applications to larger problems (e.g., datasets with hundreds of millions of cells). We document speed improvements of up to 1800%. We also demonstrate scaling of problem sizes up to 437 million grid cells. These improvements allow modelers to work with higher resolution data, larger landscapes and perform sensitivity analysis effortlessly. These improvements accelerate the pace of innovation, helping modelers address pressing challenges like species range shifts under climate change. Our collaboration between ecologists and computer scientists has led to the use of connectivity models to inform conservation decisions. Further, these next generation connectivity models will produce results faster, facilitating stronger engagement with decision-makers.
{"title":"Circuitscape in Julia: High Performance Connectivity Modelling to Support Conservation Decisions","authors":"Ranjan Anantharaman, K. Hall, Viral B. Shah, A. Edelman","doi":"10.21105/jcon.00058","DOIUrl":"https://doi.org/10.21105/jcon.00058","url":null,"abstract":"Connectivity across landscapes influences a wide range of conservation-relevant ecological processes, including species movements, gene flow, and the spread of wildfire, pests, and diseases. Recent improvements in remote sensing data suggest great potential to advance connectivity models, but computational constraints hinder these advances. To address this challenge, we upgraded the widely-used Circuitscape connectivity package to the high performance Julia programming language. Circuitscape.jl allows users to solve problems faster via improved parallel processing and solvers, and supports applications to larger problems (e.g., datasets with hundreds of millions of cells). We document speed improvements of up to 1800%. We also demonstrate scaling of problem sizes up to 437 million grid cells. These improvements allow modelers to work with higher resolution data, larger landscapes and perform sensitivity analysis effortlessly. These improvements accelerate the pace of innovation, helping modelers address pressing challenges like species range shifts under climate change. Our collaboration between ecologists and computer scientists has led to the use of connectivity models to inform conservation decisions. Further, these next generation connectivity models will produce results faster, facilitating stronger engagement with decision-makers.","PeriodicalId":443465,"journal":{"name":"JuliaCon Proceedings","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-06-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130291076","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：