Building an Application-Specific Grid Computing Environment Using ITBL for Nuclear Material Engineering

Effective use of nuclear fuel is an important issue in nuclear material engineering. Pu recycle is refocused for effective use of nuclear fuel. MOX and inert matrix fuels (e.g., ZrO₂-PuO₂) are expected for effective burning Pu, however, Pu material is difficult to manage due to its radiotoxicity. As a result, cost of experimental facilities is very expensive. As computer simulation not only cuts cost for experimental researches but also provides deep understandings in atomic behavior, we have performed molecular dynamics (MD) simulations to research its material characteristics. In order to obtain realistic results, many atoms and many time steps are essential, however, such computation takes too much long time. So a parallelized program code is executed on a grid computing environment provided by an Information Technology Based Laboratory (ITBL) project. Its grid computing infrastructure (hereafter an ITBL system) provides users a seamless computing environment and many kinds of software tools such as a file manager, a program execution manager, and a cooperative tool for AVS/Express on a grid computing environment. Furthermore, a client application program interface (API) is provided to build a variety of grid applications on a client terminal PC for accessing the ITBL’s functionalities. As there is a strong request from users in our material simulation research to utilize their native visualization software, we have selected the client API to build an application-specific grid computing environment which cooperates with the visualization software. We have built a customized graphical user interface (GUI) computing environment on a client terminal PC by using the Java client API. It provides a seamless access to ITBL’s computational resources from a user’s terminal PC. It also assists choosing parameters for the computation in parameter survey runs. Moreover, it enables successive processing of computation on remote parallel computers and visualization on a user’s terminal PC in a single operation. The environment also provides a user friendly GUI interface for parameter controls and monitoring of submitted jobs. As a result, this computing environment removes difficulties in manual operations for parallel computations and visualizations in parameter survey runs. So, it prevents users from mistakes in the operations. This environment is expected to accelerate finding procedure for good nuclear fuel.