The Computational Materials Repository

More documents

Recommendations

Info

56 Computational Materials RepositoryThis is done by relaxing the atoms (relaxing means that the positions of theatoms are optimized so that sum of the forces acting on each atom is minimal).In order to save computation time, all atoms but the hydrogen ones were fixedduring the relaxation. The best stable compounds were then completely relaxedin order to see whether the postulated structure was the correct one or if therewas a different one. Finally the compounds that allow to bind hydrogen, butnot too strongly were selected as candidates and the weight percent of hydrogenwas calculated. An other example is to answer the question whether or not anadsorbate would ”dock” on a surface or not. In this case the surface and a fewlayers are modeled and the adsorbate placed on different locations on the surface.Upon comparing the DFT energies, the most favored position can be identifiedand it can be determined whether the adsorbate would stay. There are manyother properties apart from the energy and the updated positions that can becalculated with modern electronic structure codes. An extensive introduction toelectronic structure calculations and the theory behind is for example providedby [26].This chapter starts by explaining the challenges that were faced when designingCMR. In the next step an overview of the system that we built is presented,followed by the description of each component. Then possible improvments areaddressed in the outlook and the project is later compared with other implementations.Without giving away too much details it may be helpful to know that thecore of CMR is implemented in Python, is able to use a MySQL database andcan provides access to the data with a web server that is programmed in PHP.The database and web server are optional for a minimal system 4.3.1. CMR isusually used on a Linux system.3.2 CMR ChallengesAt the beginning there was the idea to collect data from different electronicstructure simulators, and analyze them with VMDF (paper [5]) or an other tool.This was at the very beginning of my PhD. Very soon after, the CAMD SummerSchool 2008 took place where the researchers participated in a computationalmaterials design project. The generated data had to be collected and analyzed.We were a small team that designed the workflow and I was able to gain invaluableexperience by implementing the system to collect and analyze the data.The project and the implementation is outlined in paper [25].When designing the Computational Repository I had to find compromisesbetween ideas, design wishes, features, resources, performance and prospects. Inthis section I would like to discuss the faced challenges when building CMR. Thefirst challenge was to figure out what the expected features would be. BecauseCMR should handle many different kinds of problems I decided to create it
3.3 System Components and Processes 57generic with a focus on electronic structure simulators. This has the advantagethat it can solve many different problems, but might not be optimal for some ofthem.The requirements for CMR were to aggregate, store, monitor, analyze, presentand share data from electronic structure simulators. Every of these requirementimposes its own challenges: aggregation requires that CMR can read the outputfrom different codes (3.3.4.1), storing must assure that the collected data canbe read also years later even if the original program is unavailable (3.3.3.2),monitoring should provide an almost instant view of the collected data, and theanalysis options must satisfy power users that prefer to work with programmableinterfaces (3.3.1.2) and casual users that favor more convenient access methods,like a web interface (3.3.1.3). Sharing can be performed through the interfaces(3.3.1) or by sending db-files.In order to be able to analyze data it was decided that the user should beallowed to enhance the data with custom fields, keywords, and attach scripts andrelevant files to provide more background information. Another challenge relatedto the analysis is the wish to be able to show what data was used to derivedresults (e.g. which calculations were involved when calculating the chemisorptionenergy). (2.7).In terms of usability CMR must not depend on not too many third-partyproducts and provide most of it’s functionality out of the box. For exampleCMR must run independently of a database and a web-server which is especiallyimportant when CMR is executed on a computer cluster or on a user’s computerwith limited access to resources or no database access.More important requirements are to be able to assure data consistency andprevent data loss, even if processes crash, errors occur or invalid data is tried tobe uploaded (3.3.4.2). Another important issue was to find ways to provide thedata with an acceptable delay.The biggest challenge though was to provide a system that is complete andprovides the expected functionality - or at least can be extended to provide it.This might seem trivial, but the way from a concept to a working implementationrequires to handle a lot of technicalities, special cases and technical limitations.3.3 System Components and ProcessesThe data flow of an CMR installation for an institute was already briefly presentedin chapter 2. The focus in this section lies on providing more informationabout the components, show how they interact and explain their purpose. Thedata flow diagram in Fig. 3.1 highlights the three main type of components: theuser interfaces (blue) that are used to interact with the data, the repositories(purple), that store the data, and the db-files (red font) that are containers fortransferring and storing data. The arrows denote interactions/processes betweenthe components.The conversion (3.3.4.1)) to db-files (3.3.3.2) is a basic functionality of the
Page 3:
Document HistoryThis document bases
Page 8 and 9: 8 Contents
Page 10 and 11: 10 CONTENTS3.3.2.3 CMR Database . .
Page 12 and 13: 12 Introductionis an integral part
Page 14 and 15: 14 IntroductionFigure 1.2: The PHP/
Page 16 and 17: 16 Introduction
Page 18 and 19: 18 Introduction and usage of CMRAll
Page 20 and 21: 20 Introduction and usage of CMR•
Page 22 and 23: 22 Introduction and usage of CMRimp
Page 24 and 25: 24 Introduction and usage of CMRFig
Page 26 and 27: 26 Introduction and usage of CMRAt
Page 34 and 35: 34 Introduction and usage of CMR1 T
Page 36 and 37: 36 Introduction and usage of CMRloo
Page 38 and 39: 38 Introduction and usage of CMRto
Page 40 and 41: 40 Introduction and usage of CMRAAf
Page 42 and 43: 42 Introduction and usage of CMRTo
Page 44 and 45: 44 Introduction and usage of CMRmem
Page 46 and 47: 46 Introduction and usage of CMRato
Page 54 and 55: 54 Introduction and usage of CMR
Page 58 and 59: 58 Computational Materials Reposito
Page 80 and 81: 80 Appendixputer system, Date, HF,
Page 82 and 83: 82 Appendix4.2 PHPUI script to cont
Page 84 and 85: 84 Appendix4.3 Deployment Examples
Page 86 and 87: 86 Appendix4.4 Inside a db-fileA mi
Page 88 and 89: 88 Bibliography
Page 90 and 91: 90 BIBLIOGRAPHY[11] Anubhav Jain, G
Page 92 and 93: 92 BIBLIOGRAPHY[35] XML Technology.
show all

The Computational Materials Repository

Create successful ePaper yourself

Delete template?

Save as template?