Research And Implementation Of A Data Management System For Visualization In Scientific Computing

The aim of large scale scientific computing visualization is to process data and produce results conversion graphic or visual images, this technology has become an indispensable means of scientific research. With the development of high-performance computing technology, scientific data scale and quantity increases sharply, so it is an important problem of how to organize and access these massive data effectively. With the boost of large scale scientific computing visualization technology, the requirement of network based remote visualization technology is urgent, and remote visualization technology becomes a hot research topic.Firstly, an organization method of data information is designed and implemented. It employed a combined method to manage raw data using file system and database techniques, which increases the access efficiency to memory.Secondly, a new conception, visual metadata and a design of automatic metadata mining and extracting are proposed. According to this design, we defined a general rule library for metadata mining and extracting, and a processing function library for data visualization. Then we develop two modules: one for real-time tracking of raw data, and the other for data visualization.Thirdly, an architecture of web-based data management system in visualization based on B/S model is proposed. And based on this architecture, an application system is realized.Finally, we test our system using particle simulation program. Experiments show that this system can track data result in real time, automatically generate metadata summaries, and efficiently visualize data.