Some Key Issues On The HPC Based Spatial Data Application Platform Of Multi-resolution

In the dissertation, some key problems about spatial data application were discussed. A unified digital platform based on HPC was proposed to develop a digital virtual 3-d coordinate geometry background for various applications.The achievements and novels of the dissertation include:1. A parallelization structure for the platform was proposed with mobile agents technology. It adopted parallel computing algorithms and constructed with 3 blocks of user's interfacing, net resourcing and managing center. The dissertation emphasized the managing center block which is stacked with 4 levels: task response, load balancing, database managing and computing hardware.2. A concept of parallel computing entropy is introduced and defined. Based on which, an algorithm was developed to balance the loads among computing nodes. In the dissertation, task amount, rather than task number was employed to survey the load dispatched to nodes. The relationship between the entropy and the program executing time was evaluated. The result indicates that the two parameters are highly correlated.3. A method was presented to describe and optimize the parallelization process of DOM and DEM. The optimization of the process was discussed in the dissertation theoretically.4. The complexity of the algorithm was analyzed and some experiments are carried out on SC-21C HPC for both kinds of the data to demonstrate the effect of the algorithms.5. As an application of the above investigates, a problem from practice called the optimization for spatial horizontal overcastting net is defined and studied based on GA in the last chapter. The load balancing algorithm based on parallel computing entropy is employed to speed the optimization process.