Research On High Throughput Automatic Calculation Of High Efficient Catalytic Material Prediction Based On LSF

Molecular sieves are a class of crystalline materials with homogeneous microporous channels that have been widely used in various fields since they have been discovered.The functional properties of molecular sieves are inextricably linked to their skeletal structures.However,the structure of the molecular sieves has so far been only 232,which is far from satisfying the growing industrial demand.With the development of computer science and molecular engineering,it is feasible to predict the theoretical molecular sieve structure by computer method under the guidance of molecular engineering.There are many molecular sieve prediction methods in the world,including the Fra Gen prediction method which was developed by State Key Laboratory of Inorganic Synthesis & Preparative Chemistry Jilin University and can predict more reasonable molecular structures under the same conditions.However,this method has many problems: 1,This method can not simultaneously deal with different molecular structure,2,This method requires a lot of human intervention during the prediction process,3,Partial operators in prediction phase take a lot of time,4,The process of data storage is cumbersome.Aiming at resolving the above problems,this paper designed a high-throughput and automated computing architecture for catalytic material prediction based on LSF.This paper also realized the high-throughput parallel design of the chemical process and 11 automation modules,which enables the system to predict the molecular structure automatically and effectively,so that free people from the cumbersome data processing and process convergence.And this paper also designed and implemented 2 system resource maintenance module: orphan job detection module and impossible task detection module,used to maintain the system queue resources and CPU,memory resources.In addition,this article also provides users with the function of automatic database entry,the query interface of the virtual molecular structure and user operations real-time monitoring interface,allowing users to more efficiently monitor their own operations.This work is support by the National key research and development program of China — “The high throughput prediection,preparation and application of the efficient catalytic materials”.The realization of high-throughput and automated catalytic material prediction system based LSF can predict molecular sieve structure automatically and save the human resources costed by data processing.Moreover,the high-throughput calculation of the molecular structure predicted can be used to submit large quantities of tasks at one time,to complete the calculation of massive data efficiently,to increase the reasonable virtual molecular structure generated per unit time,to improve the throughput and save time cost.